Compare commits
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
6edcabf777 | ||
|
|
787984afe5 | ||
|
|
836d68c354 | ||
|
|
abce8465f8 | ||
|
|
3f0d5e87fe | ||
|
|
0bb82e4115 |
No files matched your search
+38
-5
@@ -81,11 +81,13 @@ if(MSVC AND CMAKE_GENERATOR MATCHES "Ninja" AND NOT CMAKE_CL_SHOWINCLUDES_PREFIX
|
||||
"messages in the language CMake probes for.")
|
||||
endif()
|
||||
|
||||
# Verbatim copies of the upstream native library; see THIRD_PARTY_NOTICES.md.
|
||||
# These files are never edited: they are the validated DSP kernels.
|
||||
# Sources copied verbatim from the upstream native library; see
|
||||
# THIRD_PARTY_NOTICES.md. binaural_renderer.cpp is still an unmodified copy and
|
||||
# is never edited; the unchanged reused headers include/eac3joc_core.h and
|
||||
# src/joc_bitstream/joc_huffman_tables.h belong to the same set. The JOC core,
|
||||
# the speaker renderer and the remaining tables started as copies and are
|
||||
# maintained in-tree now, so they are listed as internal sources below.
|
||||
set(JOC_REUSED_SOURCES
|
||||
src/joc_core/eac3joc_core.cpp
|
||||
src/speaker/speaker_renderer.cpp
|
||||
src/binaural/binaural_renderer.cpp
|
||||
)
|
||||
|
||||
@@ -94,6 +96,9 @@ set(JOC_INTERNAL_SOURCES
|
||||
src/adm/adm_tracks.cpp
|
||||
src/binaural/jochrtf_runtime.cpp
|
||||
src/binaural/jochrtf_renderer.cpp
|
||||
src/binaural/multichannel_layout.cpp
|
||||
src/binaural/pcm_resampler.cpp
|
||||
src/virtual_speaker/virtual_speaker.cpp
|
||||
src/eac3_transport/eac3_reader.cpp
|
||||
src/emdf/emdf_parser.cpp
|
||||
src/foundation/bit_reader.cpp
|
||||
@@ -114,13 +119,18 @@ set(JOC_INTERNAL_SOURCES
|
||||
src/io/inflate.cpp
|
||||
src/io/npy.cpp
|
||||
src/io/npy_writer.cpp
|
||||
src/io/wav_pcm.cpp
|
||||
src/io/process.cpp
|
||||
src/io/wav_writer.cpp
|
||||
src/io/zip_reader.cpp
|
||||
src/joc_bitstream/bed_layout_map.cpp
|
||||
src/joc_bitstream/joc_parser.cpp
|
||||
src/joc_core/eac3joc_core.cpp
|
||||
src/joc_core/objects16.cpp
|
||||
src/oamd/oamd_math.cpp
|
||||
src/oamd/oamd_params.cpp
|
||||
src/oamd/oamd_parser.cpp
|
||||
src/speaker/speaker_renderer.cpp
|
||||
src/speaker/speaker_layout_lookup.cpp
|
||||
src/speaker/speaker_step.cpp
|
||||
src/speaker/speaker_trim.cpp
|
||||
@@ -236,6 +246,23 @@ if(JOC_BUILD_TESTS OR JOC_BUILD_DEVTOOLS)
|
||||
PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/src"
|
||||
)
|
||||
target_link_libraries(joc_core_impl PUBLIC Threads::Threads)
|
||||
# The same sources back both the shared library and this static archive. In
|
||||
# the archive the C entry points are plain definitions: EJOC_STATIC stops the
|
||||
# kernel sources that carry their own EJOC_BUILD_DLL define from marking them
|
||||
# as dllexports, which would collide with the DLL import library in a link
|
||||
# that uses both libraries (the tests do).
|
||||
target_compile_definitions(joc_core_impl PUBLIC EJOC_STATIC)
|
||||
endif()
|
||||
|
||||
if(JOC_BUILD_TESTS OR JOC_BUILD_DEVTOOLS)
|
||||
add_executable(joc_multichannel src/cli/joc_multichannel.cpp)
|
||||
target_link_libraries(joc_multichannel PRIVATE joc_core_impl)
|
||||
target_include_directories(joc_multichannel PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/src")
|
||||
set_target_properties(joc_multichannel PROPERTIES
|
||||
MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>")
|
||||
if(WIN32)
|
||||
target_link_libraries(joc_multichannel PRIVATE shell32)
|
||||
endif()
|
||||
endif()
|
||||
|
||||
add_library(joc_core SHARED
|
||||
@@ -291,6 +318,10 @@ if(JOC_BUILD_TESTS)
|
||||
target_link_libraries(joc_rosella_filterbank_tests PRIVATE joc_core_impl)
|
||||
target_include_directories(joc_rosella_filterbank_tests PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/src")
|
||||
add_test(NAME rosella_filterbank COMMAND joc_rosella_filterbank_tests)
|
||||
add_executable(joc_oamd_metadata_tests tests/test_oamd_metadata.cpp)
|
||||
target_link_libraries(joc_oamd_metadata_tests PRIVATE joc_core_impl)
|
||||
target_include_directories(joc_oamd_metadata_tests PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/src")
|
||||
add_test(NAME oamd_metadata COMMAND joc_oamd_metadata_tests)
|
||||
# The public headers must stay valid C and C++: these targets exist to prove it.
|
||||
add_library(joc_headers_c OBJECT tests/test_headers.c)
|
||||
target_include_directories(joc_headers_c PRIVATE "${CMAKE_CURRENT_SOURCE_DIR}/include")
|
||||
@@ -317,7 +348,8 @@ if(JOC_BUILD_TESTS OR JOC_BUILD_DEVTOOLS)
|
||||
list(APPEND JOC_MSVC_TARGETS joc_core_impl)
|
||||
endif()
|
||||
if(JOC_BUILD_TESTS)
|
||||
list(APPEND JOC_MSVC_TARGETS joc_tests joc_rosella_filterbank_tests joc_headers_c joc_headers_cpp)
|
||||
list(APPEND JOC_MSVC_TARGETS joc_tests joc_rosella_filterbank_tests
|
||||
joc_oamd_metadata_tests joc_headers_c joc_headers_cpp)
|
||||
endif()
|
||||
if(JOC_BUILD_DEVTOOLS)
|
||||
endif()
|
||||
@@ -358,6 +390,7 @@ target_sources(joc_core PRIVATE
|
||||
include/eac3joc_core.h
|
||||
src/joc_bitstream/joc_huffman_tables.h
|
||||
src/joc_core/qmf_tables.h
|
||||
src/oamd/oamd_parser.h
|
||||
src/speaker/speaker_layouts.h
|
||||
)
|
||||
|
||||
|
||||
+92
-12
@@ -113,7 +113,7 @@ This build adds `--bed` (pre-decoded
|
||||
|
||||
## Speaker layouts and channel order
|
||||
|
||||
`--speaker-layout` accepts the eleven layouts below, and the PCM/WAV channel order is the order in
|
||||
`--speaker-layout` accepts the twelve layouts below, and the PCM/WAV channel order is the order in
|
||||
the table:
|
||||
|
||||
| Layout | Channel order |
|
||||
@@ -122,6 +122,7 @@ the table:
|
||||
| 3.1 | L, R, C, LFE |
|
||||
| 4.0 | L, R, Lrs, Rrs |
|
||||
| 5.1 | L, R, C, LFE, Ls, Rs |
|
||||
| 6.0 | L, R, Ls, Rs, Lrs, Rrs |
|
||||
| 7.1 | L, R, C, LFE, Ls, Rs, Lrs, Rrs |
|
||||
| 5.1.2 | L, R, C, LFE, Ls, Rs, Ltm, Rtm |
|
||||
| 5.1.4 | L, R, C, LFE, Ls, Rs, Ltf, Rtf, Ltr, Rtr |
|
||||
@@ -140,7 +141,7 @@ into mask-bit order. Players, ASIO, DAWs and other devices have to map the chann
|
||||
themselves; this project guarantees the content and numbering of each PCM channel, not that a
|
||||
downstream player recognises Dolby speaker positions on its own.
|
||||
|
||||
`--speaker-layout` accepts only the eleven names in the table; anything else fails as an unknown
|
||||
`--speaker-layout` accepts only the twelve names in the table; anything else fails as an unknown
|
||||
layout.
|
||||
|
||||
Speaker rendering reads warp, trim and front/back balance from the OAMD metadata; the CLI has no
|
||||
@@ -149,6 +150,18 @@ then fold back to 5.1.x with PLIIx or Standard (Lo/Ro) coefficients, and calibra
|
||||
follows the final target layout. Source trajectories and binaural rendering do not pre-apply
|
||||
these controls; see [mathematics §12](docs/math.en.md) for the parameters and evaluation order.
|
||||
|
||||
Speaker rendering also applies the OAMD object attributes: per-block gain/inactive,
|
||||
zone/elevation, size/snap/divergence, the priority size budget, element 5 position
|
||||
refinement, ISF basis matrices, and bed distribution. ADM export and the binaural
|
||||
paths still place point objects and do not apply these speaker target coefficients.
|
||||
|
||||
Object position updates are scheduled from the complete `frame_state`: speaker and
|
||||
binaural rendering both accept multiple position blocks and programmes with more than
|
||||
16 objects and schedule every block at its own time and ramp. The binaural audio input
|
||||
is still one LFE plus fifteen object sources; the extra position tracks are followed
|
||||
but not panned. ADM export keeps the 15-track single-block position view and rejects
|
||||
anything larger as an unsupported variant.
|
||||
|
||||
## Library integration
|
||||
|
||||
The engine and the file task are exposed by `joc_core.h`: `joc_task_validate` /
|
||||
@@ -191,18 +204,49 @@ tests/ unit tests (CTest, self-contained, no external data)
|
||||
docs/ mathematics, the binaural rendering flow and SIMD dispatch
|
||||
```
|
||||
|
||||
Eight files under `src/` are byte-identical copies of the upstream JustOneCacophony
|
||||
native library and are never edited (see [THIRD_PARTY_NOTICES.md](THIRD_PARTY_NOTICES.md)).
|
||||
The QMF and binaural tables and the JOC Huffman tables under `src/` are written from
|
||||
publicly documented standards; [THIRD_PARTY_NOTICES.md](THIRD_PARTY_NOTICES.md) records
|
||||
their sources.
|
||||
|
||||
The ISF compressed base coefficients and layout descriptors are obtained from
|
||||
lookup tables stored in the generated header (`src/oamd/isf_tables.h`), which is
|
||||
compiled into the binary with the code, so nothing reads a data file at run
|
||||
time. Gain, size, priority, position, divergence, delta and refinement values are
|
||||
computed in code, while the size kernel is generated from the speaker layout
|
||||
region geometry and bed routing is derived from those layouts.
|
||||
|
||||
## Mathematical implementation
|
||||
|
||||
The main documented stages are:
|
||||
|
||||
- dense JOC differential reconstruction and dequantization;
|
||||
- parameter-band mapping to 64 QMF subbands;
|
||||
- cross-frame parameter interpolation (hard cut when no previous state exists, zero-matrix publishing after an object disappears);
|
||||
- the frame-level JOC state machine (normal, hold, hard cut, core passthrough);
|
||||
- analysis/inverse QMF, surround delay, and FIR state;
|
||||
- the 1217-sample LFE delay with its sample-wise clipgain ramp;
|
||||
- OAMD Q15 coordinate conversion;
|
||||
- equal-power panning over target-layout regions;
|
||||
- OAMD metadata application (per-block gain and inactive, zone/elevation, size kernel, snap, divergence, position refinement, priority size budget, ISF matrices, bed distribution);
|
||||
- layout-dependent position compensation and sample-wise gain ramps;
|
||||
- float32 and PCM24 output quantization;
|
||||
- Rosella and compiled `.jochrtf` binaural rendering, with exactly-once delay/phase and special LFE.
|
||||
|
||||
See the [mathematical notes](docs/math.en.md) for the equations used by the decoding
|
||||
and rendering process.
|
||||
|
||||
## Compatibility note
|
||||
|
||||
### How `joc_clipgain` is applied
|
||||
|
||||
`joc_clipgain` is folded into the **analysis-side scalar**: `scalar = 1/16 x g` (`g = 1 + (y/32)·2^(x-4)`,
|
||||
from the 8-bit field `x:3 | y:5`). The x16 at the output cancels the 1/16, so the net effect is a multiply
|
||||
by `g` and the clamp therefore sits **after** the gain (`clamp(16·IQMF·g_interp)`). The scalar is kept in
|
||||
two per-frame slots: the first 256 samples of a frame ramp linearly from the previous value, the rest stays
|
||||
constant, and the first frame after a reset takes the new value directly. For `g = 1` the interpolation is the identity and `clamp(x)·1 == clamp(x·1)`.
|
||||
`joc_clipgain` is folded into the **analysis-side scalar**: `scalar = 1/16 x g`, with
|
||||
`g = 2^(x-4)·(1 + y/32)` from the 8-bit field `x:3 | y:5`; the range is 0.0625..15.75. The x16 at the
|
||||
output cancels the 1/16, so the net effect is a multiply by `g` and the clamp therefore sits **after**
|
||||
the gain (`clamp(16·IQMF·g_interp)`). The scalar is kept in two per-frame slots: the first 256 samples
|
||||
of a frame ramp linearly from the previous value, the rest stays constant, and the first frame after a
|
||||
reset takes the new value directly. For `g = 1` the interpolation is the identity and
|
||||
`clamp(x)·1 == clamp(x·1)`. LFE bypasses the analysis but applies the same full `g` as a sample-wise
|
||||
ramp before its 1217-sample delay.
|
||||
|
||||
`object_delay_samples` defaults to **1473**, preserving the behaviour of the existing
|
||||
implementation; it is a configurable field and changing it changes the OAMD/ADM time
|
||||
@@ -239,9 +283,45 @@ only a unique best score is used and the others are recorded as a diagnostic
|
||||
$k\in\{2,4\}\to8/9$) is recorded, but bed channels are not output channels under the current
|
||||
contract.
|
||||
|
||||
An audio frame with no JOC EMDF at all (for example when the encoder wrote none for the first
|
||||
frame) emits silent object channels while the QMF analysis and synthesis state keep advancing,
|
||||
and it is counted separately in the statistics.
|
||||
### Frame status and the JOC decode state machine
|
||||
|
||||
Every frame carries a status; a non-zero status is reported as a diagnostic. A non-zero status
|
||||
makes the frame invalid for rendering, so the renderer holds the previous matrices or passes the
|
||||
core PCM through instead of decoding that frame's matrix: 2 unsupported downmix configuration,
|
||||
3 object count clamped to 15, 4 unknown parameter-band index, 5 invalid `offset_ts`, 6 trailing
|
||||
data after byte padding, and 7 an ID14 payload that ends inside `joc_data`. A file task refuses a
|
||||
`joc_dmx_config_idx` outside 0..4 earlier, in the metadata scan, as an unsupported variant.
|
||||
Status 3 still yields the clamped object list for diagnostics, status 2 and 4 stop the parse at
|
||||
the failing header field, and status 7 keeps the complete header, clipgain, sequence counter and
|
||||
object parameters, with only the object matrices missing.
|
||||
|
||||
Status 7 is classified only when the EMDF container itself is complete: a container length past
|
||||
the syncframe, or a payload too short for the 12-bit `joc_header`, stays a transport error, and
|
||||
the file task and stream API additionally require the container's ID11 payload to parse. A
|
||||
complete container whose payload ends inside `joc_data` takes the invalid-frame path (the
|
||||
previous matrices are held), so decoding continues instead of stopping; the stream statistics
|
||||
count those frames in `truncated_joc_frames`, and `joc_check_id14_padding` still reports
|
||||
`JOC_ERR_BITSTREAM_TRUNCATED` for them.
|
||||
|
||||
Rendering then runs a per-frame state machine: normal, hold, hard cut, and core passthrough. An
|
||||
unusable-JOC frame that arrives during normal rendering enters hold, where the previous frame-end
|
||||
matrices keep being published; six consecutive unusable frames move the internal hold state on but
|
||||
stay in hold, and a valid frame resynchronises the sequence number. A sequence number of 0 after
|
||||
the renderer left its initial state, a sequence mismatch, or a change of the declared core channel
|
||||
count is a hard cut: every object matrix is zeroed and each object's previous state becomes
|
||||
the zero matrix, while a core channel count change also clears the LFE delay ring but keeps
|
||||
the clipgain ramp state. Before the first valid frame the renderer passes the core PCM
|
||||
through.
|
||||
|
||||
A frame with no JOC EMDF at all (for example when the encoder wrote none for the first frame)
|
||||
always takes the passthrough path: the eight core planes L, R, C, LFE, Ls, Rs, (Lrs, Rrs) are
|
||||
written to output slots 0..7 by identity, each delayed by 1217 samples, slots 8..15 stay zero,
|
||||
and the 8-plane history keeps advancing. Those frames are counted separately in the statistics,
|
||||
and E-AC-3 input without JOC metadata (ID11 + ID14) is outside the supported scope. A frame whose
|
||||
ID14 payload ended inside `joc_data` is not one of these: it takes the hold path above and is
|
||||
counted separately in the stream statistics (`truncated_joc_frames`). While hold or
|
||||
passthrough runs without fresh parameters, the analysis scalar and the LFE ramp keep the
|
||||
currently held clipgain.
|
||||
|
||||
## License
|
||||
|
||||
|
||||
@@ -99,7 +99,7 @@ joc_cli input.eac3 --metadata-only --print-metadata summary # 只解析并打
|
||||
|
||||
## 扬声器布局与声道顺序
|
||||
|
||||
`--speaker-layout` 只接受下表 11 个布局,PCM/WAV 的声道顺序就是表里的顺序:
|
||||
`--speaker-layout` 只接受下表 12 个布局,PCM/WAV 的声道顺序就是表里的顺序:
|
||||
|
||||
| 布局 | 声道顺序 |
|
||||
|---|---|
|
||||
@@ -107,6 +107,7 @@ joc_cli input.eac3 --metadata-only --print-metadata summary # 只解析并打
|
||||
| 3.1 | L, R, C, LFE |
|
||||
| 4.0 | L, R, Lrs, Rrs |
|
||||
| 5.1 | L, R, C, LFE, Ls, Rs |
|
||||
| 6.0 | L, R, Ls, Rs, Lrs, Rrs |
|
||||
| 7.1 | L, R, C, LFE, Ls, Rs, Lrs, Rrs |
|
||||
| 5.1.2 | L, R, C, LFE, Ls, Rs, Ltm, Rtm |
|
||||
| 5.1.4 | L, R, C, LFE, Ls, Rs, Ltf, Rtf, Ltr, Rtr |
|
||||
@@ -122,13 +123,21 @@ joc_cli input.eac3 --metadata-only --print-metadata summary # 只解析并打
|
||||
PCM 声道顺序严格按上表,不按掩码位序重排。播放器、ASIO、DAW 或其他设备需要按声道编号自行完成
|
||||
输出映射;本项目保证各 PCM 声道的内容和编号符合上表,不保证下游播放器自动识别杜比扬声器位置。
|
||||
|
||||
`--speaker-layout` 只接受上表 11 个名字;其它名字按未知布局报错。
|
||||
`--speaker-layout` 只接受上表 12 个名字;其它名字按未知布局报错。
|
||||
|
||||
扬声器渲染从 OAMD 元数据读取 warp、trim 和 front/back balance;命令行不提供 warp 覆盖。
|
||||
warp 0/1 直接渲染;warp 2/3 先渲染到 7.1.x 中间布局,再按 PLIIx 或 Standard(Lo/Ro)
|
||||
折回 5.1.x,校准参数始终按最终目标布局选择。原始对象轨迹与双耳渲染不预处理这些控制,
|
||||
参数与求值顺序见[数学说明 §12](docs/math.md)。
|
||||
|
||||
扬声器渲染还会应用 OAMD 的对象属性:每块增益/inactive、zone/elevation、size/snap/divergence、
|
||||
priority 尺寸预算、element 5 位置细化、ISF 基矩阵和 bed distribute。ADM 导出与双耳路径仍按点
|
||||
对象放置,不应用这些扬声器目标系数。
|
||||
|
||||
对象位置的调度按完整 `frame_state` 进行:扬声器与双耳渲染都接受多 position block 和超过 16
|
||||
个对象的节目,按各块的时间与 ramp 逐个调度;双耳渲染的音频输入仍是 1 LFE + 15 对象,多出的
|
||||
位置轨道只跟踪不参与声像。ADM 导出仍使用 15 轨、单 block 的对象位置视图,超出时按变体报错。
|
||||
|
||||
## 库集成
|
||||
|
||||
引擎与文件任务使用 `joc_core.h`:`joc_task_validate` / `joc_task_execute` 跑一个文件任务并
|
||||
@@ -164,17 +173,41 @@ tests/ 单元测试(CTest,自足,不需要外部素材)
|
||||
docs/ 数学说明、双耳渲染流程与 SIMD 派发
|
||||
```
|
||||
|
||||
`src/` 下 8 个文件是 JustOneCacophony 原生库的逐字节副本,永不修改(见
|
||||
[THIRD_PARTY_NOTICES.md](THIRD_PARTY_NOTICES.md))。
|
||||
`src/` 下的 QMF/双耳表与 JOC Huffman 表按公开标准实现;来源记录见
|
||||
[THIRD_PARTY_NOTICES.md](THIRD_PARTY_NOTICES.md)。
|
||||
|
||||
ISF 压缩基系数与布局描述以查表方式提供:表数据写在生成头文件 `src/oamd/isf_tables.h` 中,
|
||||
与代码一起编译进二进制,运行时不读取数据文件。gain、size、priority、位置、divergence
|
||||
与差分/细化值都在代码中计算,size 核由扬声器布局的 region 几何生成,bed 路由由布局表推导。
|
||||
|
||||
## 数学实现
|
||||
|
||||
主要文档化阶段:
|
||||
|
||||
- dense JOC 差分还原与去量化;
|
||||
- 参数带到 64 个 QMF 子带的映射;
|
||||
- 跨帧参数插值(无上一帧时硬切,对象消失后按零矩阵发布);
|
||||
- 帧级 JOC 状态机(正常 / 保持 / 硬切 / 核心直通);
|
||||
- analysis / inverse QMF、环绕声道延迟与 FIR 状态;
|
||||
- LFE 1217-sample 延迟与逐点 clipgain 斜坡;
|
||||
- OAMD Q15 坐标转换;
|
||||
- 基于目标布局 region 的等功率声像;
|
||||
- OAMD 元数据应用(每块增益与 inactive、zone/elevation、size 核、snap、divergence、位置细化、priority 尺寸预算、ISF 矩阵、bed 分配);
|
||||
- 布局位置补偿与逐样本增益斜坡;
|
||||
- float32 与 PCM24 输出量化;
|
||||
- Rosella 与编译好的 `.jochrtf` 双耳渲染,含 exactly-once delay/phase 与 special LFE。
|
||||
|
||||
解码与渲染所用的公式见[数学说明](docs/math.md)。
|
||||
|
||||
## 兼容性说明
|
||||
|
||||
### `joc_clipgain` 的施加方式
|
||||
|
||||
`joc_clipgain` 在**分析端**并入分析标量:`标量 = 1/16 × g`(`g = 1 + (y/32)·2^(x−4)`,8 bit 字段
|
||||
`x:3 | y:5`)。出口的 ×16 抵消 1/16,净效果是 `×g`,因此 clamp 处在增益**之后**
|
||||
`joc_clipgain` 在**分析端**并入分析标量:`标量 = 1/16 × g`(`g = 2^(x−4)·(1 + y/32)`,8 bit 字段
|
||||
`x:3 | y:5`,值域 0.0625..15.75)。出口的 ×16 抵消 1/16,净效果是 `×g`,因此 clamp 处在增益**之后**
|
||||
(`clamp(16·IQMF·g_interp)`)。标量按帧保存在两个槽里:本帧前 256 个采样从上一帧值线性过渡、之后为
|
||||
常数,复位后的第一帧直接取新值。`g = 1` 时插值恒等,且 `clamp(x)·1 == clamp(x·1)`。
|
||||
常数,复位后的第一帧直接取新值。`g = 1` 时插值恒等,且 `clamp(x)·1 == clamp(x·1)`。LFE 走独立
|
||||
通路,但使用同一个完整 `g` 的逐点斜坡,并乘在 1217-sample 延迟之前。
|
||||
|
||||
`object_delay_samples` 默认 **1473**,与既有实现的行为保持一致;它是可配置字段,改动它会
|
||||
改变 OAMD/ADM 时间对齐。上游调查认为该值应为 0,本项目为保持逐字节等价暂不改默认值。
|
||||
@@ -202,8 +235,32 @@ OAMD 与 JOC 载荷 groupid 相同且 ID14 能完整解析的优先,其次只
|
||||
其余记入 diagnostics(`emdf/extra_container_ignored`)。额外那一对在 10 槽 ADM 骨架上的落位
|
||||
($k=1\to6/7$、$k\in\{2,4\}\to8/9$)会被记录,但床通道按现行契约不作为输出通道。
|
||||
|
||||
音频帧内完全没有 JOC EMDF 时(例如编码器没写首帧元数据),该帧的对象通道静音,
|
||||
QMF 分析与合成状态照常推进,并在统计里单独计数。
|
||||
### 帧状态与 JOC 解码状态机
|
||||
|
||||
每个帧都带一个状态,非零状态按诊断报告。非零状态使该帧不能用于渲染,渲染器因此保持上一帧矩阵或输出
|
||||
核心 PCM 直通,而不还原该帧的矩阵:2 不支持的下混配置,3 对象数截断到 15,4 未知参数带索引,
|
||||
5 非法 `offset_ts`,6 字节对齐后存在尾随数据,7 ID14 载荷在 `joc_data` 内提前结束。文件任务对
|
||||
`joc_dmx_config_idx` 超出 0..4 的输入更早在元数据扫描阶段按不支持变体拒绝。状态 3 仍给出截断后的
|
||||
对象列表(供诊断),状态 2、4 在出错的头字段处停止解析,状态 7 则保留完整的头、clipgain、序号和
|
||||
对象参数,只缺对象矩阵。
|
||||
|
||||
状态 7 只在 EMDF 容器本身完整时才分类:容器长度越界、或载荷短到连 12 bit 的 `joc_header` 都放
|
||||
不下仍是传输错误,文件任务与流接口还要求同一容器里的 ID11 载荷能解析。容器完整而载荷在
|
||||
`joc_data` 中提前结束的帧按无效帧处理(保持上一帧矩阵),文件任务和流接口因此继续解码而不中断,
|
||||
流接口统计里的 `truncated_joc_frames` 记录这类帧,`joc_check_id14_padding` 对它们仍报
|
||||
`JOC_ERR_BITSTREAM_TRUNCATED`。
|
||||
|
||||
渲染随后按帧级状态机运行:正常、保持、硬切、核心直通。正常渲染期间出现 JOC 参数不可用的帧时
|
||||
进入保持,继续发布上一帧末的矩阵;连续六个不可用帧只把内部保持状态推进一档、仍保持;出现有效
|
||||
帧后重新对齐序号。渲染器离开初始状态后序号回到 0、序号不匹配、或声明的核心声道数变化都
|
||||
触发硬切:所有对象矩阵清零、每个对象的 previous 状态置零,核心声道数变化还清空 LFE 延迟环
|
||||
但保留 clipgain 斜坡状态。首个有效帧之前,渲染器输出核心 PCM 直通。
|
||||
|
||||
音频帧内完全没有 JOC EMDF 时(例如编码器没写首帧元数据),该帧一律走直通路径:8 路核心平面
|
||||
L、R、C、LFE、Ls、Rs、(Lrs, Rrs) 按 identity 写入输出槽 0..7,每路延迟 1217 采样,槽 8..15
|
||||
保持零,8 平面历史照常推进。这类帧在统计里单独计数;不含 JOC 元数据(ID11 + ID14)的 E-AC-3
|
||||
输入不在支持范围。ID14 载荷在 `joc_data` 内提前结束的帧不属于这一类:它走上面的保持路径,
|
||||
在流接口统计里单独计数(`truncated_joc_frames`)。保持或直通期间没有新参数时,分析标量与 LFE 斜坡沿用当前保持的 clipgain。
|
||||
|
||||
## 许可
|
||||
|
||||
|
||||
+6
-8
@@ -26,7 +26,7 @@ performs no parsing or conversion of measurement data formats.
|
||||
```powershell
|
||||
joc_cli input.m4a --binaural `
|
||||
--compiled-hrtf-cache path\to\subject.jochrtf `
|
||||
--kernels data\rosella_kernels.npz `
|
||||
--kernels path\to\rosella_kernels.npz `
|
||||
--binaural-mode mid --binaural-tail-seconds 5.0
|
||||
```
|
||||
|
||||
@@ -95,8 +95,8 @@ compiler does not sample 77 FFT bins. Defaults are `1e-3` projection ridge and
|
||||
`1e-5` SH ridge. Coincident directions are merged before a spherical-Voronoi
|
||||
weighted ridge fit.
|
||||
|
||||
The fixed resource is `data/rosella_kernels.npz`, which implements publicly
|
||||
standardized filter banks, computable from the following formulas.
|
||||
The fixed tables implement publicly standardized filter banks, compile into the
|
||||
sources, and are computable from the following formulas.
|
||||
|
||||
The hybrid analysis kernels are defined in [3GPP TS 26.405 / ETSI TS 126 405](https://www.etsi.org/deliver/etsi_ts/126400_126499/126405/06.00.00_60/ts_126405v060000p.pdf),
|
||||
Section 5.2.2 (Table 1 $Q=8$/
|
||||
@@ -125,11 +125,9 @@ $8+4+4$ child partition):
|
||||
|
||||
$$Y_p = \sum_{q\in C_p}\Bigl(\mathrm{Re}X_q + j\,s_q\,\mathrm{Im}X_q\Bigr),\qquad s_q\in\{\pm1\}$$
|
||||
|
||||
The loader verifies the archive and every array by SHA-256; the table version,
|
||||
all array hashes, and the 77 reference band-center values are part of the cache
|
||||
key. Public availability of a standard does not by itself grant permission to
|
||||
practice related patent claims. See
|
||||
[`data/README.en.md`](../data/README.en.md) and
|
||||
The tables are fixed at compile time, so no external archive is loaded and
|
||||
nothing enters a cache key at run time. Public availability of a standard does
|
||||
not by itself grant permission to practice related patent claims. See
|
||||
[`THIRD_PARTY_NOTICES.md`](../THIRD_PARTY_NOTICES.md) for the sources and the
|
||||
rights boundary.
|
||||
|
||||
|
||||
+4
-5
@@ -24,7 +24,7 @@ cache,不是交换格式。
|
||||
```powershell
|
||||
joc_cli input.m4a --binaural `
|
||||
--compiled-hrtf-cache path\to\subject.jochrtf `
|
||||
--kernels data\rosella_kernels.npz `
|
||||
--kernels path\to\rosella_kernels.npz `
|
||||
--jochrtf-room on `
|
||||
--binaural-mode mid --binaural-tail-seconds 5.0
|
||||
joc_cli input.m4a --binaural --personalized-headphone HRTF\binaural.personalized_headphone
|
||||
@@ -78,7 +78,7 @@ joc_cli input.m4a --binaural --personalized-headphone HRTF\binaural.personalized
|
||||
ridge 为 `1e-3`,SH ridge 为 `1e-5`。同方向 measurement 先合并,再用球面 Voronoi
|
||||
面积权重做 ridge fit。
|
||||
|
||||
固定表位于 `data/rosella_kernels.npz`,实现公开标准化的滤波器组,各表可由如下
|
||||
固定表按公开标准化的滤波器组实现并编译进源码,各表可由如下
|
||||
公式计算。
|
||||
|
||||
hybrid 分析核定义于 [3GPP TS 26.405 / ETSI TS 126 405](https://www.etsi.org/deliver/etsi_ts/126400_126499/126405/06.00.00_60/ts_126405v060000p.pdf)
|
||||
@@ -106,9 +106,8 @@ $8+4+4$ 子带划分):
|
||||
|
||||
$$Y_p = \sum_{q\in C_p}\Bigl(\mathrm{Re}X_q + j\,s_q\,\mathrm{Im}X_q\Bigr),\qquad s_q\in\{\pm1\}$$
|
||||
|
||||
loader 校验 archive 和每个数组的 SHA-256;table version、所有数组 hash 与
|
||||
77 个 band-center 参考值都属于 cache key。标准可公开获取不等于获准实施相关
|
||||
专利;更多来源信息见 [`data/README.md`](../data/README.md) 与
|
||||
固定表在编译期确定,运行时不加载外部 archive,也不参与缓存键。标准可公开
|
||||
获取不等于获准实施相关专利;来源与权利边界见
|
||||
[`THIRD_PARTY_NOTICES.md`](../THIRD_PARTY_NOTICES.md)。
|
||||
|
||||
## `.jochrtf`
|
||||
|
||||
+95
-13
@@ -147,15 +147,17 @@ If the clipgain field consists of integer $x$ and mantissa $y$, then
|
||||
|
||||
$$
|
||||
G_{\mathrm{clip}}=
|
||||
1+\frac{y}{32}2^{x-4}.
|
||||
2^{x-4}\left(1+\frac{y}{32}\right).
|
||||
$$
|
||||
|
||||
It is folded into the analysis-side scalar $S=\frac{1}{16}G_{\mathrm{clip}}$, which multiplies
|
||||
every core-bed sample before the QMF analysis; the $\times16$ after the object inverse QMF cancels the
|
||||
$1/16$, so the net effect is a multiply by $G_{\mathrm{clip}}$ and the $[-1,1]$ clamp sits **after** the
|
||||
gain. The scalar keeps two per-frame slots: the first 256 samples of a frame ramp linearly from the
|
||||
previous frame's value to this frame's value and the rest stays constant, and the first frame after a
|
||||
reset takes the new value directly. LFE bypasses this scalar.
|
||||
Its range is $[0.0625,15.75]$; at $x=4$ it reduces to $1+y/32$. It is folded into the analysis-side
|
||||
scalar $S=\frac{1}{16}G_{\mathrm{clip}}$, which multiplies every core-bed sample before the QMF
|
||||
analysis; the $\times16$ after the object inverse QMF cancels the $1/16$, so the net effect is a multiply
|
||||
by $G_{\mathrm{clip}}$ and the $[-1,1]$ clamp sits **after** the gain. The scalar keeps two per-frame
|
||||
slots: the first 256 samples of a frame ramp linearly from the previous frame's value to this frame's
|
||||
value and the rest stays constant, and the first frame after a reset takes the new value directly. LFE
|
||||
uses a separate path but applies the same full $G_{\mathrm{clip}}$ as a sample-wise ramp before its
|
||||
1217-sample delay; see section 8.
|
||||
|
||||
## 3. Parameter-band expansion and time interpolation
|
||||
|
||||
@@ -190,7 +192,13 @@ Thus $p(b)=k$ if and only if $b\in\mathcal B_k$. Other parameter-band counts use
|
||||
|
||||
### 3.2 One-data-point interpolation
|
||||
|
||||
Let $P_{o,c,b}$ be the previous frame-end value and $D_{o,c,p(b)}$ the current target. For slot $t=0\ldots23$:
|
||||
Let $P_{o,c,b}$ be the previous frame-end value and $D_{o,c,p(b)}$ the current target. When the object
|
||||
has no previous state yet (the first frame of the stream, or the object's first appearance), a single
|
||||
data point keeps all 24 slots at $D_{o,c,p(b)}$ (hard cut, no fade-in from zero); with two data points,
|
||||
the first 12 slots stay at $D_0$ and the last 12 slots still interpolate from $D_0$ to $D_1$ using the
|
||||
second segment of section 3.3 (with $P$ set to $D_0$).
|
||||
|
||||
With a previous state, for slot $t=0\ldots23$:
|
||||
|
||||
$$
|
||||
\alpha_t=\frac{t+1}{24},
|
||||
@@ -208,11 +216,31 @@ $$
|
||||
M_{o,c,b,23}=D_{o,c,p(b)}.
|
||||
$$
|
||||
|
||||
This value then becomes the previous state for the next frame.
|
||||
This value then becomes the previous state for the next frame. Once an object's present flag goes from 1
|
||||
to 0, it keeps being published with an all-zero matrix and its previous state is updated to zero until it
|
||||
becomes present again.
|
||||
|
||||
### 3.3 Multiple data points
|
||||
|
||||
When a frame contains two data points, `offset_ts` gives the segment boundary. Each segment uses the same linear relation between the previous and next targets; step mode switches targets at the designated slot.
|
||||
`offset_ts` gives the first slot at which each data point becomes active, using the raw 5-bit value
|
||||
(legal range $0\ldots23$); the values of one object must be strictly increasing. In the default mode
|
||||
(`slope_idx=0`), two data points $D_0$ and $D_1$ interpolate as
|
||||
|
||||
$$
|
||||
M_{o,c,b,t}=
|
||||
\begin{cases}
|
||||
(1-\alpha_t)P_{o,c,b}+\alpha_tD_{0}, & t=0\ldots10,\ \alpha_t=\dfrac{t+1}{12},\\[1mm]
|
||||
D_{0}, & t=11,\\[1mm]
|
||||
(1-\beta_t)D_{0}+\beta_tD_{1}, & t=12\ldots23,\ \beta_t=\dfrac{t-11}{12}.
|
||||
\end{cases}
|
||||
$$
|
||||
|
||||
When no previous state exists, the same formulas apply with $P_{o,c,b}$ set to $D_0$.
|
||||
|
||||
Step mode (`slope_idx=1`) switches targets directly at the designated slots: for $t<o_0$ it keeps the
|
||||
previous frame-end value $P_{o,c,b}$ (or $D_0$ when no previous state exists), for $o_0\le t<o_1$ it
|
||||
uses $D_0$, and for $t\ge o_1$ it uses $D_1$; with a single data point it keeps $P$ for $t<o_0$ and uses
|
||||
$D_0$ from $o_0$ on.
|
||||
|
||||
## 4. Analysis QMF for core PCM
|
||||
|
||||
@@ -337,15 +365,21 @@ where $r=0\ldots63$. Synthesis state must advance continuously by slot.
|
||||
|
||||
## 8. LFE path
|
||||
|
||||
LFE bypasses the object matrix and inverse QMF and uses a 1217-sample delay. After the input and output scale factors cancel:
|
||||
LFE bypasses the object matrix and inverse QMF. The core LFE samples are first multiplied by the
|
||||
sample-wise gain ramp $G[n]$ (the same full $G_{\mathrm{clip}}$ as section 2.4: for $n<256$,
|
||||
$G[n]=G_{\mathrm{old}}+n(G_{\mathrm{new}}-G_{\mathrm{old}})/256$; for $n\ge256$, $G[n]=G_{\mathrm{new}}$;
|
||||
on the first frame or after a reset, $G_{\mathrm{old}}=G_{\mathrm{new}}$). The scaled samples then
|
||||
enter the 1217-sample delay, and the read-out is only clamped, with no further gain:
|
||||
|
||||
$$
|
||||
y_{\mathrm{LFE}}[n]=
|
||||
\mathrm{clip}\left(
|
||||
x_{\mathrm{LFE,core}}[n-1217],-1,1
|
||||
x_{\mathrm{LFE,core}}[n-1217]\cdot G[n-1217],-1,1
|
||||
\right).
|
||||
$$
|
||||
|
||||
Here $G[k]$ uses the frame-local position of input sample $k$, i.e. $k\bmod1536$.
|
||||
|
||||
## 9. OAMD coordinates
|
||||
|
||||
The lateral and longitudinal grids use $N=62$; the height grid uses $N=15$. The quantizer is
|
||||
@@ -358,7 +392,7 @@ q_N(k)=
|
||||
\right).
|
||||
$$
|
||||
|
||||
OAR coordinates are
|
||||
The normalized room coordinates are
|
||||
|
||||
$$
|
||||
u=\frac{q_1}{32768},
|
||||
@@ -800,3 +834,51 @@ $$
|
||||
- Multiple OAMD position blocks must be scheduled in time order.
|
||||
- A limiter is separate post-processing and is not included in the mixing equations above.
|
||||
- Trim and front-back balance affect speaker rendering only; binaural rendering does not read them.
|
||||
|
||||
## 16. OAMD metadata application (speaker path)
|
||||
|
||||
When mapping object PCM to the target layout, speaker rendering applies the OAMD
|
||||
object attributes in addition to the geometric panning:
|
||||
|
||||
- per-block integer gain `Qg` (unity 4096) and `inactive`; the gain is applied to
|
||||
the target coefficients once, and the LFE uses its owning metadata object's
|
||||
gain/active state;
|
||||
- full basic/render status 0/1/2/3 block expansion, same-object previous-block
|
||||
inheritance, and local mask updates;
|
||||
- `zone_constraints_idx` region selection mapped through `[0,1,2,3,4,5,6,0]`;
|
||||
`enable_elevation=0` forces z=0 before position compensation and size spread;
|
||||
- `object_size_idx` 5-bit sizes, the `size_curve` piecewise mapping, and the size
|
||||
kernel; the kernel is generated at runtime from region geometry with
|
||||
equal-power grouped panning, the S448 weight and a power sum (20/20/8
|
||||
virtual points, 20 size nodes, 35/4 position nodes; boundary rows
|
||||
X(0), X(1), Y(0), Y(1), Z(1));
|
||||
- `snap` first-argmax speaker snapping, `divergence` left/right point-source
|
||||
combination, and element 5 high-precision position refinement;
|
||||
- `default_priority`/`priority_bits` choose which size objects keep their size
|
||||
when the size budget is limited;
|
||||
- `isf_idx` basis matrices (signed coefficients, no per-column normalization)
|
||||
with the per-basis compensation;
|
||||
- `b_bed_chan_distribute` label candidates, weight normalization, missing-channel
|
||||
fallback, and the direct/distribute compensation.
|
||||
|
||||
The ISF path orders the N ISF object audio essences into the column vector t
|
||||
according to the M.U.L.Z order of `isf_config`, selects the input configuration
|
||||
by `isf_idx` and the matrix M for the target output layout, and evaluates
|
||||
y = M·t: each ISF channel is distributed over the speakers by the corresponding
|
||||
matrix row, LFE bypasses the matrix, and each basis carries its own
|
||||
compensation. The matrix coefficients and layout descriptors are obtained from
|
||||
lookup tables stored in the compiled-in `src/oamd/isf_tables.h`; nothing reads a
|
||||
data file at run time.
|
||||
|
||||
Gain, size, priority, position, divergence, delta and refinement values are
|
||||
computed in code (`src/oamd/oamd_parser.*`) from formulas or fixed constants; the
|
||||
size kernel (`src/oamd/oamd_math.*`) is generated at runtime from region geometry
|
||||
with grouped equal-power panning, the S448 weight and a power sum. Bed labels,
|
||||
routing and reference positions are derived from the speaker layout tables
|
||||
(`src/speaker/speaker_layouts.h`) by
|
||||
`src/joc_bitstream/bed_layout_map.{h,cpp}`.
|
||||
ADM export and the binaural paths still use object positions only and do not apply
|
||||
these speaker target coefficients. The binaural path schedules positions from the
|
||||
complete `frame_state`: multiple position blocks and programmes with more than 16
|
||||
objects are scheduled block by block at their own time and ramp, and its position
|
||||
track list is not limited to 16; ADM export keeps the 15-track single-block view.
|
||||
+73
-9
@@ -147,13 +147,15 @@ $$
|
||||
|
||||
$$
|
||||
G_{\mathrm{clip}}=
|
||||
1+\frac{y}{32}2^{x-4}.
|
||||
2^{x-4}\left(1+\frac{y}{32}\right).
|
||||
$$
|
||||
|
||||
它并入分析端标量 $S=\frac{1}{16}G_{\mathrm{clip}}$,乘在核心 bed 的每个采样上(QMF 分析之前);
|
||||
值域为 $[0.0625,15.75]$;$x=4$ 时化为 $1+y/32$。它并入分析端标量
|
||||
$S=\frac{1}{16}G_{\mathrm{clip}}$,乘在核心 bed 的每个采样上(QMF 分析之前);
|
||||
对象 inverse QMF 之后的 $\times16$ 抵消 $1/16$,净效果是乘 $G_{\mathrm{clip}}$,因此钳位 $[-1,1]$
|
||||
处在增益**之后**。标量按帧保留两个槽,本帧前 256 个采样从上一帧值线性过渡到本帧值、之后为常数,
|
||||
复位后的第一帧直接取新值。LFE 走旁通通路,不参与该标量。
|
||||
复位后的第一帧直接取新值。LFE 走独立通路,但使用同一个完整 $G_{\mathrm{clip}}$ 的逐点斜坡:
|
||||
斜坡在进入 1217-sample 延迟之前乘到 LFE 采样上,见 §8。
|
||||
|
||||
## 3. 参数带展开与时间插值
|
||||
|
||||
@@ -188,7 +190,12 @@ $$
|
||||
|
||||
### 3.2 单数据点插值
|
||||
|
||||
令上一帧末值为 $P_{o,c,b}$,当前目标值为 $D_{o,c,p(b)}$。对时槽 $t=0\ldots23$:
|
||||
令上一帧末值为 $P_{o,c,b}$,当前目标值为 $D_{o,c,p(b)}$。对象没有 previous 状态时
|
||||
(整条流第一帧或对象首次出现):单数据点情形 24 个时槽恒为 $D_{o,c,p(b)}$(硬切,
|
||||
不从零淡入);双数据点情形前 12 个时槽为 $D_0$,后 12 个时槽按 §3.3 的第二段公式
|
||||
从 $D_0$ 插值到 $D_1$(把该节的 $P$ 取为 $D_0$)。
|
||||
|
||||
有 previous 状态时,对时槽 $t=0\ldots23$:
|
||||
|
||||
$$
|
||||
\alpha_t=\frac{t+1}{24},
|
||||
@@ -206,11 +213,28 @@ $$
|
||||
M_{o,c,b,23}=D_{o,c,p(b)}.
|
||||
$$
|
||||
|
||||
该值随后成为下一帧的 previous 状态。
|
||||
该值随后成为下一帧的 previous 状态。对象 present 由 1 变 0 后,从该帧起继续按全零矩阵发布,
|
||||
并把 previous 更新为零矩阵,直到它重新 present=1。
|
||||
|
||||
### 3.3 多数据点
|
||||
|
||||
当一帧含两个数据点时,`offset_ts` 给出分段边界。每一段在上一目标和下一目标之间使用相同的线性关系;阶跃模式则在指定时槽直接切换目标。
|
||||
`offset_ts` 给出各数据点开始生效的时隙序号,取原始 5 bit 值(合法范围 $0\ldots23$);
|
||||
同一对象的数据点必须严格递增。默认模式(`slope_idx=0`)下,两个数据点 $D_0,D_1$ 的插值为
|
||||
|
||||
$$
|
||||
M_{o,c,b,t}=
|
||||
\begin{cases}
|
||||
(1-\alpha_t)P_{o,c,b}+\alpha_tD_{0}, & t=0\ldots10,\ \alpha_t=\dfrac{t+1}{12},\\[1mm]
|
||||
D_{0}, & t=11,\\[1mm]
|
||||
(1-\beta_t)D_{0}+\beta_tD_{1}, & t=12\ldots23,\ \beta_t=\dfrac{t-11}{12}.
|
||||
\end{cases}
|
||||
$$
|
||||
|
||||
没有 previous 状态时同样使用上面的公式,只是把 $P_{o,c,b}$ 取为 $D_0$。
|
||||
|
||||
阶跃模式(`slope_idx=1`)在指定时槽直接切换目标:$t<o_0$ 保持上一帧末值 $P_{o,c,b}$
|
||||
(没有 previous 时取 $D_0$),$o_0\le t<o_1$ 取 $D_0$,$t\ge o_1$ 取 $D_1$;
|
||||
单数据点时 $t<o_0$ 保持 $P$、$t\ge o_0$ 取 $D_0$。
|
||||
|
||||
## 4. 核心 PCM 的 analysis QMF
|
||||
|
||||
@@ -334,15 +358,20 @@ $$
|
||||
|
||||
## 8. LFE 路径
|
||||
|
||||
LFE 不经过对象矩阵或 inverse QMF,而是使用 1217-sample 延迟。输入与输出端的比例因子抵消后:
|
||||
LFE 不经过对象矩阵或 inverse QMF。核心 LFE 先乘逐点增益斜坡 $G[n]$(与 §2.4 同一个完整
|
||||
$G_{\mathrm{clip}}$:$n<256$ 时 $G[n]=G_{\mathrm{old}}+n(G_{\mathrm{new}}-G_{\mathrm{old}})/256$,
|
||||
$n\ge256$ 时 $G[n]=G_{\mathrm{new}}$,首帧或复位后 $G_{\mathrm{old}}=G_{\mathrm{new}}$),
|
||||
再进入 1217-sample 延迟;读出后只做钳位,不再乘任何增益:
|
||||
|
||||
$$
|
||||
y_{\mathrm{LFE}}[n]=
|
||||
\mathrm{clip}\left(
|
||||
x_{\mathrm{LFE,core}}[n-1217],-1,1
|
||||
x_{\mathrm{LFE,core}}[n-1217]\cdot G[n-1217],-1,1
|
||||
\right).
|
||||
$$
|
||||
|
||||
其中 $G[k]$ 取输入采样 $k$ 所在帧内的位置,即 $k\bmod1536$。
|
||||
|
||||
## 9. OAMD 坐标
|
||||
|
||||
横向和纵向网格使用 $N=62$,高度网格使用 $N=15$。量化函数为
|
||||
@@ -355,7 +384,7 @@ q_N(k)=
|
||||
\right).
|
||||
$$
|
||||
|
||||
OAR 坐标为
|
||||
归一化房间坐标为
|
||||
|
||||
$$
|
||||
u=\frac{q_1}{32768},
|
||||
@@ -774,3 +803,38 @@ $$
|
||||
- 多个 OAMD position block 必须按其时间顺序调度。
|
||||
- limiter 属于独立后处理,不包含在上述混音公式中。
|
||||
- trim 与前后平衡只作用于扬声器渲染;双耳渲染不读取它们。
|
||||
|
||||
## 16. OAMD 元数据应用(扬声器路径)
|
||||
|
||||
扬声器渲染把对象 PCM 映射到目标布局时,除几何声像外还应用 OAMD 的对象属性:
|
||||
|
||||
- 每块整数增益 `Qg`(单位增益 4096)与 `inactive`;增益在生成目标系数时乘一次,
|
||||
LFE 使用所属元数据对象的增益/活动状态;
|
||||
- basic/render 状态 0/1/2/3 的完整块展开、同对象前一块继承与局部 mask 更新;
|
||||
- `zone_constraints_idx` 选择区域,映射为 `[0,1,2,3,4,5,6,0]`;`enable_elevation=0`
|
||||
在位置补偿和尺寸扩散前令 z=0;
|
||||
- `object_size_idx` 的 5 bit 尺寸、`size_curve` 分段线性映射与尺寸扩散核;
|
||||
尺寸核按分组等功率声像、S448 权重和幂和,在运行时用区域几何生成:
|
||||
20/20/8 个虚拟点、20 个尺寸节点、35/4 个位置节点,边界表为
|
||||
X(0)、X(1)、Y(0)、Y(1)、Z(1);
|
||||
- `snap` 的首个最大系数扬声器吸附、`divergence` 的左右两点合成、
|
||||
element 5 的高精度位置补充;
|
||||
- `default_priority`/`priority_bits` 在 size 数量预算不足时决定保留尺寸的对象;
|
||||
- `isf_idx` 的 ISF 基矩阵(可含负系数,不做逐列归一化)与逐 basis 补偿;
|
||||
- `b_bed_chan_distribute` 的 bed 标签候选筛选、权重归一化、缺失声道 fallback 和
|
||||
直达/distribute 的补偿。
|
||||
|
||||
ISF 路径把 N 路 ISF object audio essences 按 `isf_config` 的 M.U.L.Z 顺序组成
|
||||
列向量 t,按 `isf_idx` 选输入配置、按目标输出布局选矩阵 M,再计算 y = M·t:
|
||||
每个 ISF 声道按矩阵对应的一行分配到目标扬声器,LFE 不参与矩阵、直接旁路,
|
||||
每个 basis 另有补偿。矩阵系数与布局描述通过查表获得,表数据写在编译进源码的
|
||||
`src/oamd/isf_tables.h` 中,运行时不读取数据文件。
|
||||
|
||||
gain、size、priority、位置、divergence 与差分/细化值由 `src/oamd/oamd_parser.*`
|
||||
按公式或固定常量计算,size 扩散核由 `src/oamd/oamd_math.*` 从区域几何生成
|
||||
(分组等功率声像 + S448 权重 + 幂和)。bed 标签、路由和参考位置由
|
||||
`src/joc_bitstream/bed_layout_map.*` 从 `src/speaker/speaker_layouts.h` 的布局表
|
||||
推导。
|
||||
ADM 导出与双耳路径仍只使用对象位置,不应用这些扬声器目标系数。双耳路径按完整
|
||||
`frame_state` 调度位置:多 position block 与超过 16 个对象的节目都按各块的时间与 ramp
|
||||
逐个调度,双耳位置轨道数不限于 16;ADM 导出保持 15 轨单 block 视图。
|
||||
+81
-4
@@ -3,19 +3,22 @@
|
||||
#include <stdint.h>
|
||||
|
||||
/* EJOC_STATIC: this branch exists for building the sources directly into an application, where nothing is imported or exported. */
|
||||
#if defined(_WIN32)
|
||||
#define EJOC_CALL __cdecl
|
||||
#else
|
||||
#define EJOC_CALL
|
||||
#endif
|
||||
|
||||
#if defined(EJOC_STATIC)
|
||||
#define EJOC_API
|
||||
#define EJOC_CALL __cdecl
|
||||
#elif defined(_WIN32)
|
||||
#if defined(EJOC_BUILD_DLL)
|
||||
#define EJOC_API __declspec(dllexport)
|
||||
#else
|
||||
#define EJOC_API __declspec(dllimport)
|
||||
#endif
|
||||
#define EJOC_CALL __cdecl
|
||||
#else
|
||||
#define EJOC_API __attribute__((visibility("default")))
|
||||
#define EJOC_CALL
|
||||
#endif
|
||||
|
||||
#ifdef __cplusplus
|
||||
@@ -23,13 +26,15 @@ extern "C" {
|
||||
#endif
|
||||
|
||||
enum {
|
||||
EJOC_ABI_VERSION = 3,
|
||||
EJOC_ABI_VERSION = 4,
|
||||
EJOC_FRAME_SAMPLES = 1536,
|
||||
EJOC_TIMESLOTS = 24,
|
||||
EJOC_SUBBANDS = 64,
|
||||
EJOC_CORE_CHANNELS = 5, /* downmix configurations 0 and 3 */
|
||||
EJOC_MAX_CORE_CHANNELS = 7, /* downmix configurations 1, 2 and 4 */
|
||||
EJOC_OUTPUT_CHANNELS = 16,
|
||||
EJOC_CORE_OUT_CHANNELS = 8, /* passthrough order: L,R,C,LFE,Ls,Rs,(Lrs,Rrs) */
|
||||
EJOC_LFE_SLOT = 3, /* LFE slot of the passthrough order */
|
||||
EJOC_MAX_OBJECTS = 15,
|
||||
EJOC_MAX_DPOINTS = 2,
|
||||
EJOC_MAX_PARAMETER_BANDS = 23,
|
||||
@@ -43,6 +48,21 @@ enum {
|
||||
EJOC_BINAURAL_HYBRID_BANDS = 77
|
||||
};
|
||||
|
||||
/* Per-frame status of ejoc_renderer_render_frame(). A status outside 0..6 is
|
||||
* the renderer's own no-metadata marker; the parse-status values match
|
||||
* joc_frame_params.status (JOC_FRAME_STATUS_* in joc_core.h). Any non-zero
|
||||
* status makes the frame invalid for matrix rendering: the renderer holds the
|
||||
* previous matrices or passes the core PCM through instead of decoding it. */
|
||||
enum {
|
||||
EJOC_FRAME_STATUS_OK = 0,
|
||||
EJOC_FRAME_STATUS_UNSUPPORTED_DMX = 2,
|
||||
EJOC_FRAME_STATUS_OBJECTS_CLAMPED = 3,
|
||||
EJOC_FRAME_STATUS_BAD_PARAMETER_BAND = 4,
|
||||
EJOC_FRAME_STATUS_BAD_OFFSET_TS = 5,
|
||||
EJOC_FRAME_STATUS_TRAILING_DATA = 6,
|
||||
EJOC_FRAME_STATUS_NO_METADATA = 8
|
||||
};
|
||||
|
||||
typedef void* ejoc_renderer_handle;
|
||||
typedef void* ejoc_speaker_renderer_handle;
|
||||
typedef void* ejoc_binaural_renderer_handle;
|
||||
@@ -103,6 +123,63 @@ EJOC_API int EJOC_CALL ejoc_renderer_process(
|
||||
float output_scale,
|
||||
float* output16_planar);
|
||||
|
||||
/*
|
||||
Frame-level entry point: one E-AC-3 audio frame per call, with the seq state
|
||||
machine and the per-object previous-matrix state kept inside the renderer. This
|
||||
is what produces the 16-channel object result for the ADM, speaker and binaural
|
||||
backends; ejoc_renderer_process() stays as a single-frame kernel for A/B tests.
|
||||
|
||||
Fixed array layouts, identical to ejoc_renderer_process():
|
||||
bed_planar [core_channel_count][1536] (JOC lane order L,R,C,Ls,Rs,(pair))
|
||||
lfe [1536] or NULL (zeros; the LFE delay ring still advances)
|
||||
n_bands, n_dpoints, slope_idx [15]
|
||||
offset_ts [15][2] (raw 5-bit JOC offsets, 0..23)
|
||||
dq [15][2][7][23]
|
||||
output16 [16][1536] (ch0 = LFE, ch1..15 = objects)
|
||||
|
||||
frame_status is a JOC_FRAME_STATUS_* / EJOC_FRAME_STATUS_* value. has_metadata
|
||||
is 0 for an audio frame without a JOC EMDF container (frame_status must then be
|
||||
EJOC_FRAME_STATUS_NO_METADATA); such a frame always passes the 8 core planes
|
||||
through to output slots 0..7, delayed by 1217 samples, with slots 8..15 zero.
|
||||
declared_core_channels is the JOC header's n_channels (0 when unknown); a declared
|
||||
change against the previous frame's actual count turns the frame into a hard cut
|
||||
with the LFE ring cleared. phase_shift is sigma(k) = (k in {3,4}) for the frame's
|
||||
downmix configuration; phase_shift_valid = 0 keeps the previous switch (an
|
||||
unsupported downmix configuration has no usable one). clipgain is the frame's
|
||||
joc_clipgain when frame_status == 0. seq is the 10-bit JOC sequence counter.
|
||||
When has_metadata is 0 the object descriptors are ignored.
|
||||
|
||||
Slot 0 of output16 is the LFE, slots 1..15 are the 15 objects; each frame's
|
||||
values are float32(clamped PCM) * output_scale, exactly like the reference.
|
||||
*/
|
||||
typedef struct ejoc_frame_input {
|
||||
uint32_t struct_size;
|
||||
uint32_t struct_version;
|
||||
const float* bed_planar;
|
||||
const float* lfe;
|
||||
uint32_t core_channel_count;
|
||||
uint32_t declared_core_channels;
|
||||
uint32_t phase_shift;
|
||||
uint32_t phase_shift_valid;
|
||||
uint32_t frame_status;
|
||||
uint32_t seq;
|
||||
uint32_t has_metadata;
|
||||
uint32_t object_mask;
|
||||
const uint8_t* n_bands;
|
||||
const uint8_t* n_dpoints;
|
||||
const uint8_t* slope_idx;
|
||||
const uint8_t* offset_ts;
|
||||
const double* dq;
|
||||
double clipgain;
|
||||
float output_scale;
|
||||
float* output16_planar;
|
||||
} ejoc_frame_input;
|
||||
|
||||
enum { EJOC_FRAME_INPUT_VERSION = 1 };
|
||||
|
||||
EJOC_API int EJOC_CALL ejoc_renderer_render_frame(ejoc_renderer_handle handle,
|
||||
const ejoc_frame_input* input);
|
||||
|
||||
/*
|
||||
High-precision object-to-speaker renderer.
|
||||
|
||||
|
||||
+46
-14
@@ -36,27 +36,30 @@
|
||||
|
||||
#include "joc_version.h"
|
||||
|
||||
#define JOC_ABI_VERSION 5u
|
||||
#define JOC_FRAME_PARAMS_VERSION 1u
|
||||
#define JOC_EMDF_INFO_VERSION 1u
|
||||
#define JOC_ABI_VERSION 6u
|
||||
#define JOC_FRAME_PARAMS_VERSION 2u
|
||||
#define JOC_EMDF_INFO_VERSION 2u
|
||||
#define JOC_TASK_CONFIG_VERSION 4u
|
||||
#define JOC_TASK_RESULT_VERSION 2u
|
||||
#define JOC_EVENT_VERSION 1u
|
||||
|
||||
/* JOC_STATIC: this branch exists for building the sources directly into an application, where nothing is imported or exported. */
|
||||
#if defined(_WIN32)
|
||||
#define JOC_CALL __cdecl
|
||||
#else
|
||||
#define JOC_CALL
|
||||
#endif
|
||||
|
||||
#if defined(JOC_STATIC)
|
||||
#define JOC_API
|
||||
#define JOC_CALL __cdecl
|
||||
#elif defined(_WIN32)
|
||||
#if defined(JOC_BUILD_DLL)
|
||||
#define JOC_API __declspec(dllexport)
|
||||
#else
|
||||
#define JOC_API __declspec(dllimport)
|
||||
#endif
|
||||
#define JOC_CALL __cdecl
|
||||
#else
|
||||
#define JOC_API __attribute__((visibility("default")))
|
||||
#define JOC_CALL
|
||||
#endif
|
||||
|
||||
#ifdef __cplusplus
|
||||
@@ -152,8 +155,7 @@ JOC_API const char* JOC_CALL joc_last_error_detail(void);
|
||||
typedef struct joc_emdf_payload_info {
|
||||
uint8_t id;
|
||||
uint8_t reserved[3];
|
||||
uint16_t sample_offset; /* EMDF outer smpoffst, 11 bits */
|
||||
uint16_t reserved2;
|
||||
uint32_t sample_offset; /* EMDF outer smpoffst: variable_bits(11) */
|
||||
uint32_t bit_offset;
|
||||
uint32_t size; /* payload bytes */
|
||||
} joc_emdf_payload_info;
|
||||
@@ -170,7 +172,8 @@ typedef struct joc_emdf_info {
|
||||
|
||||
/* Per-object ID14/JOC descriptor plus its dequantized matrix.
|
||||
* dq[dp][ch][pb] is zero filled outside [0,n_dpoints) x [0,n_channels) x
|
||||
* [0,n_bands). Absent objects are entirely zero. */
|
||||
* [0,n_bands). Absent objects are entirely zero. offset_ts holds the raw
|
||||
* 5-bit JOC offsets (legal range 0..23, strictly increasing per object). */
|
||||
typedef struct joc_object_params {
|
||||
uint8_t present;
|
||||
uint8_t num_bands_idx;
|
||||
@@ -185,6 +188,26 @@ typedef struct joc_object_params {
|
||||
double dq[JOC_MAX_DPOINTS][JOC_MAX_CORE_CHANNELS][JOC_MAX_PARAMETER_BANDS];
|
||||
} joc_object_params;
|
||||
|
||||
/* Per-frame JOC parse status (joc_frame_params.status). A non-zero status makes
|
||||
* the frame invalid for rendering: the renderer holds the previous matrices or
|
||||
* passes the core PCM through instead of decoding the matrix. Statuses 3 and 7
|
||||
* still carry the complete header and object list for diagnostics; statuses 2
|
||||
* and 4 stop the parse after the failing header field, and 7 stops inside
|
||||
* joc_data, so only the per-object matrices are missing there. */
|
||||
enum {
|
||||
JOC_FRAME_STATUS_OK = 0,
|
||||
JOC_FRAME_STATUS_UNSUPPORTED_DMX = 2, /* joc_dmx_config_idx outside 0..4 */
|
||||
JOC_FRAME_STATUS_OBJECTS_CLAMPED = 3, /* declared object count above 15 */
|
||||
JOC_FRAME_STATUS_BAD_PARAMETER_BAND = 4, /* unknown num_bands index */
|
||||
JOC_FRAME_STATUS_BAD_OFFSET_TS = 5, /* offset_ts outside 0..23 or not */
|
||||
/* strictly increasing */
|
||||
JOC_FRAME_STATUS_TRAILING_DATA = 6, /* bytes after joc_data alignment */
|
||||
/* The ID14 payload ends before its joc_data is complete. The header,
|
||||
* clipgain, sequence counter and object parameters are intact, so the frame
|
||||
* stays classified and is concealed by holding the previous matrices. */
|
||||
JOC_FRAME_STATUS_TRUNCATED = 7
|
||||
};
|
||||
|
||||
typedef struct joc_frame_params {
|
||||
uint32_t struct_size;
|
||||
uint32_t struct_version;
|
||||
@@ -196,21 +219,29 @@ typedef struct joc_frame_params {
|
||||
uint8_t clipgain_x_bits;
|
||||
uint8_t clipgain_y_bits;
|
||||
uint8_t objects_clamped; /* num_objects_bits+1 exceeded 15 */
|
||||
uint32_t seq_count; /* 10-bit JOC sequence counter, parsed only */
|
||||
uint32_t seq_count; /* 10-bit JOC sequence counter */
|
||||
uint32_t present_mask; /* bit i == object i present */
|
||||
uint8_t status; /* JOC_FRAME_STATUS_* */
|
||||
uint8_t reserved[3];
|
||||
uint32_t data_end_bits; /* bit position just after joc_data */
|
||||
uint32_t trailing_bits; /* bits left after joc_data (padding/ext) */
|
||||
uint8_t tail_bytes[8]; /* first up to 8 trailing bytes, for A/B */
|
||||
double clipgain; /* 1 + (y/32) * 2^(x-4), bit-exact */
|
||||
double clipgain; /* 2^(x-4) * (1 + y/32), bit-exact */
|
||||
joc_object_params objects[JOC_MAX_OBJECTS];
|
||||
} joc_frame_params;
|
||||
|
||||
/* Parse an EMDF ID14 (JOC) payload. */
|
||||
/* Parse an EMDF ID14 (JOC) payload. A payload that ends inside joc_data is a
|
||||
* damaged frame rather than a transport error: the call succeeds, the header and
|
||||
* object parameters are filled in, and out_params->status is
|
||||
* JOC_FRAME_STATUS_TRUNCATED. A payload too short for the 12-bit joc_header is
|
||||
* unclassifiable and stays JOC_ERR_BITSTREAM_TRUNCATED. */
|
||||
JOC_API joc_error JOC_CALL joc_parse_id14(const uint8_t* payload, size_t payload_size,
|
||||
joc_frame_params* out_params);
|
||||
|
||||
/* Locate the contiguous JOC EMDF container inside one E-AC-3 syncframe and
|
||||
* parse its ID14 payload. out_emdf may be NULL. */
|
||||
* parse its ID14 payload. out_emdf may be NULL. A damaged ID14 payload is
|
||||
* reported through out_params->status like joc_parse_id14(); an incomplete
|
||||
* container stays a transport error. */
|
||||
JOC_API joc_error JOC_CALL joc_parse_eac3_frame(const uint8_t* frame, size_t frame_size,
|
||||
joc_frame_params* out_params,
|
||||
joc_emdf_info* out_emdf);
|
||||
@@ -231,7 +262,8 @@ JOC_API joc_error JOC_CALL joc_eac3_frame_bytes(const uint8_t* data, size_t size
|
||||
|
||||
/* Strict trailing-bit check for the JOC payload (A/B robustness corpus).
|
||||
* Returns JOC_ERR_BITSTREAM_PADDING when more than 7 bits are left over or the
|
||||
* leftover bits are not zero. */
|
||||
* leftover bits are not zero, and JOC_ERR_BITSTREAM_TRUNCATED when the payload
|
||||
* ends inside joc_data. */
|
||||
JOC_API joc_error JOC_CALL joc_check_id14_padding(const uint8_t* payload, size_t payload_size,
|
||||
uint32_t* out_trailing_bits);
|
||||
|
||||
|
||||
+10
-3
@@ -40,7 +40,8 @@ typedef struct joc_stream joc_stream;
|
||||
typedef enum joc_stream_input {
|
||||
JOC_STREAM_IN_EAC3 = 0, /* bare E-AC-3 syncframes (the metadata stream) */
|
||||
JOC_STREAM_IN_PCM_OBJECTS16 = 1, /* objects16 frames, planar [16][1536] each */
|
||||
JOC_STREAM_IN_CORE_PCM = 3 /* the 5.1 core PCM of the pushed E-AC-3 frames */
|
||||
JOC_STREAM_IN_CORE_PCM = 3, /* the 5.1 core PCM of the pushed E-AC-3 frames */
|
||||
JOC_STREAM_IN_PCM_LAYOUT = 4 /* interleaved 48k f32, mapped by pcm_layout */
|
||||
} joc_stream_input;
|
||||
|
||||
typedef enum joc_stream_output {
|
||||
@@ -68,6 +69,8 @@ typedef struct joc_stream_config {
|
||||
* .jochrtf field (hrtf_path) or a Rosella .personalized_headphone model. */
|
||||
const char* personalized_headphone_path; /* Rosella .personalized_headphone input */
|
||||
uint32_t jochrtf_room; /* non-zero enables the project room */
|
||||
/* PCM_LAYOUT input: "2.0", "5.1", "7.1", or NULL/"" to infer by channels. */
|
||||
const char* pcm_layout;
|
||||
} joc_stream_config;
|
||||
|
||||
typedef struct joc_stream_buffer {
|
||||
@@ -97,8 +100,9 @@ typedef struct joc_stream_status_info {
|
||||
uint64_t oamd_payloads;
|
||||
uint64_t oamd_transitions;
|
||||
/* Metadata the renderer had to work around, per audio frame: no JOC payload at
|
||||
* all, a declared object count clamped to 15, and a syncframe carrying a second
|
||||
* EMDF container that is not this frame's JOC metadata. */
|
||||
* all, an ID14 payload that ends inside joc_data (that frame holds the previous
|
||||
* matrices), a declared object count clamped to 15, and a syncframe carrying a
|
||||
* second EMDF container that is not this frame's JOC metadata. */
|
||||
uint64_t missing_joc_frames;
|
||||
uint64_t objects_clamped_frames;
|
||||
uint64_t extra_container_frames;
|
||||
@@ -108,6 +112,9 @@ typedef struct joc_stream_status_info {
|
||||
uint64_t bed_mask_mismatch_frames;
|
||||
uint32_t output_channels;
|
||||
uint32_t ended; /* 1 after flush() */
|
||||
/* Appended field: joc_stream_status() accepts the size that ends at `ended`
|
||||
* as well, so a caller built against the previous header keeps working. */
|
||||
uint64_t truncated_joc_frames;
|
||||
} joc_stream_status_info;
|
||||
|
||||
JOC_API joc_error JOC_CALL joc_stream_create(const joc_stream_config* config, joc_stream** out);
|
||||
|
||||
@@ -9,8 +9,8 @@
|
||||
#define JOC_VERSION_H
|
||||
|
||||
#define JOC_VERSION_MAJOR 1
|
||||
#define JOC_VERSION_MINOR 2
|
||||
#define JOC_VERSION_PATCH 0
|
||||
#define JOC_VERSION_STRING "1.2.0"
|
||||
#define JOC_VERSION_MINOR 3
|
||||
#define JOC_VERSION_PATCH 3
|
||||
#define JOC_VERSION_STRING "1.3.3"
|
||||
|
||||
#endif /* JOC_VERSION_H */
|
||||
@@ -363,6 +363,20 @@ std::string build_axml(const std::vector<Track>& tracks, double duration_sec, st
|
||||
out += "<position coordinate=\"Z\">" + pynum::format_fixed(block.z, 10) +
|
||||
"</position>";
|
||||
}
|
||||
const ObjectMetadata& meta = block.meta;
|
||||
if (meta.size[0] > 0.0 || meta.size[1] > 0.0 || meta.size[2] > 0.0) {
|
||||
out += "<width>" + pynum::format_fixed(meta.size[0], 6) + "</width>";
|
||||
out += "<height>" + pynum::format_fixed(meta.size[2], 6) + "</height>";
|
||||
out += "<depth>" + pynum::format_fixed(meta.size[1], 6) + "</depth>";
|
||||
}
|
||||
if (std::fabs(meta.gain - 1.0) > 1e-9) {
|
||||
out += "<gain gainUnit=\"linear\">" + pynum::format_fixed(meta.gain, 6) +
|
||||
"</gain>";
|
||||
}
|
||||
if (meta.divergence > 0.0) {
|
||||
out += "<objectDivergence positionRange=\"" +
|
||||
pynum::format_fixed(meta.divergence, 6) + "\"/>";
|
||||
}
|
||||
out += "<jumpPosition interpolationLength=\"" +
|
||||
pynum::format_fixed(static_cast<double>(block.interpolation_samples) / scale,
|
||||
5) +
|
||||
|
||||
@@ -14,6 +14,15 @@ namespace joc::adm {
|
||||
inline constexpr int kObjectCount = 15;
|
||||
inline constexpr std::uint32_t kTrackCount = 25;
|
||||
|
||||
// Per-block OAMD render metadata the ADM audioBlockFormat can express.
|
||||
struct ObjectMetadata {
|
||||
double gain = 1.0;
|
||||
// OAMD order: width, depth, height (relative units 0..1).
|
||||
double size[3] = {0.0, 0.0, 0.0};
|
||||
double divergence = 0.0;
|
||||
bool snap = false;
|
||||
};
|
||||
|
||||
struct Keyframe {
|
||||
std::int64_t rtime_samples = 0;
|
||||
double x = 0.0;
|
||||
@@ -21,6 +30,7 @@ struct Keyframe {
|
||||
double z = 0.0;
|
||||
std::int64_t duration_samples = 0;
|
||||
std::int64_t interpolation_samples = 0;
|
||||
ObjectMetadata meta;
|
||||
};
|
||||
|
||||
struct Track {
|
||||
|
||||
+61
-13
@@ -16,15 +16,22 @@ struct Point {
|
||||
double y = 0.0;
|
||||
double z = 0.0;
|
||||
std::int64_t interpolation_samples = 0;
|
||||
ObjectMetadata meta;
|
||||
};
|
||||
|
||||
bool same_meta(const ObjectMetadata& left, const ObjectMetadata& right) {
|
||||
return left.gain == right.gain && left.size[0] == right.size[0] &&
|
||||
left.size[1] == right.size[1] && left.size[2] == right.size[2] &&
|
||||
left.divergence == right.divergence && left.snap == right.snap;
|
||||
}
|
||||
|
||||
// otherwise (the reference drops silently).
|
||||
void append_point(std::vector<Point>* points, std::int64_t sample, double x, double y, double z,
|
||||
std::int64_t interpolation_samples) {
|
||||
std::int64_t interpolation_samples, const ObjectMetadata& meta) {
|
||||
if (!points->empty() && sample == points->back().sample) {
|
||||
points->back() = Point{sample, x, y, z, interpolation_samples};
|
||||
points->back() = Point{sample, x, y, z, interpolation_samples, meta};
|
||||
} else if (points->empty() || sample > points->back().sample) {
|
||||
points->push_back(Point{sample, x, y, z, interpolation_samples});
|
||||
points->push_back(Point{sample, x, y, z, interpolation_samples, meta});
|
||||
}
|
||||
}
|
||||
|
||||
@@ -52,6 +59,7 @@ void points_to_blocks(const std::vector<Point>& points, std::int64_t total_sampl
|
||||
keyframe.z = point.z;
|
||||
keyframe.duration_samples = duration;
|
||||
keyframe.interpolation_samples = std::min(point.interpolation_samples, duration);
|
||||
keyframe.meta = point.meta;
|
||||
out->push_back(keyframe);
|
||||
}
|
||||
}
|
||||
@@ -85,7 +93,8 @@ Status expand_compact(const std::vector<OamdEvent>& events, std::int64_t total_s
|
||||
|
||||
std::vector<Point> points;
|
||||
double current[3] = {events[0].x, events[0].y, events[0].z};
|
||||
append_point(&points, 0, current[0], current[1], current[2], 0);
|
||||
ObjectMetadata current_meta = events[0].meta;
|
||||
append_point(&points, 0, current[0], current[1], current[2], 0, current_meta);
|
||||
|
||||
for (std::size_t index = 1; index < events.size(); ++index) {
|
||||
const OamdEvent& event = events[index];
|
||||
@@ -131,10 +140,12 @@ Status expand_compact(const std::vector<OamdEvent>& events, std::int64_t total_s
|
||||
target[2] = z;
|
||||
interpolation = available;
|
||||
}
|
||||
append_point(&points, block_start, target[0], target[1], target[2], interpolation);
|
||||
append_point(&points, block_start, target[0], target[1], target[2], interpolation,
|
||||
event.meta);
|
||||
current[0] = event.x;
|
||||
current[1] = event.y;
|
||||
current[2] = event.z;
|
||||
current_meta = event.meta;
|
||||
}
|
||||
|
||||
points_to_blocks(points, total_samples, out);
|
||||
@@ -156,7 +167,8 @@ Status expand_dense64(const std::vector<OamdEvent>& events, std::int64_t total_s
|
||||
|
||||
std::vector<Point> points;
|
||||
double current[3] = {events[0].x, events[0].y, events[0].z};
|
||||
append_point(&points, 0, current[0], current[1], current[2], 0);
|
||||
ObjectMetadata current_meta = events[0].meta;
|
||||
append_point(&points, 0, current[0], current[1], current[2], 0, current_meta);
|
||||
|
||||
for (std::size_t index = 1; index < events.size(); ++index) {
|
||||
const OamdEvent& event = events[index];
|
||||
@@ -171,14 +183,15 @@ Status expand_dense64(const std::vector<OamdEvent>& events, std::int64_t total_s
|
||||
points.back().sample);
|
||||
}
|
||||
if (start > points.back().sample) {
|
||||
append_point(&points, start, current[0], current[1], current[2], 0);
|
||||
append_point(&points, start, current[0], current[1], current[2], 0, current_meta);
|
||||
}
|
||||
const std::int64_t ramp = std::max<std::int64_t>(0, event.ramp_samples);
|
||||
if (ramp == 0) {
|
||||
append_point(&points, start, event.x, event.y, event.z, 0);
|
||||
append_point(&points, start, event.x, event.y, event.z, 0, event.meta);
|
||||
current[0] = event.x;
|
||||
current[1] = event.y;
|
||||
current[2] = event.z;
|
||||
current_meta = event.meta;
|
||||
continue;
|
||||
}
|
||||
const std::int64_t steps = (ramp + update_block_samples - 1) / update_block_samples;
|
||||
@@ -214,12 +227,13 @@ Status expand_dense64(const std::vector<OamdEvent>& events, std::int64_t total_s
|
||||
break;
|
||||
}
|
||||
append_point(&points, sample, position[0], position[1], position[2],
|
||||
update_block_samples);
|
||||
update_block_samples, event.meta);
|
||||
future -= update_block_samples;
|
||||
}
|
||||
current[0] = event.x;
|
||||
current[1] = event.y;
|
||||
current[2] = event.z;
|
||||
current_meta = event.meta;
|
||||
}
|
||||
|
||||
points_to_blocks(points, total_samples, out);
|
||||
@@ -239,18 +253,52 @@ void TrajectoryBuilder::submit_frame(std::int64_t frame_index, const oamd::OamdU
|
||||
if (fixed_[object - 1]) {
|
||||
continue; // 固定 bed 声道:位置由布局给出,不跟 OAMD 更新
|
||||
}
|
||||
const int slot = object - 1;
|
||||
// OAMD block 0 对动态对象显式携带 gain/size/snap;divergence 只在扩展
|
||||
// 元素里出现,缺失时按 0 处理,与 speaker 渲染器的逐 payload 语义一致。
|
||||
if (update != nullptr && update->has_frame_state &&
|
||||
static_cast<std::size_t>(object) < update->frame_state.objects.size()) {
|
||||
const oamd::ObjectElementState& element =
|
||||
update->frame_state.objects[static_cast<std::size_t>(object)];
|
||||
if (!element.blocks.empty() && element.kind == oamd::ObjectKind::kDynamic) {
|
||||
const oamd::ObjectRenderState& render = element.blocks[0].render;
|
||||
ObjectMetadata& meta = meta_[slot];
|
||||
if (render.not_active) {
|
||||
meta.gain = 0.0;
|
||||
} else {
|
||||
meta.gain = static_cast<double>(render.gain_q) / 4096.0;
|
||||
if (render.size_mode == 1u || render.size_mode == 2u) {
|
||||
for (int axis = 0; axis < 3; ++axis) {
|
||||
meta.size[axis] = std::min(
|
||||
std::max(static_cast<double>(render.size_q[axis]) / 32767.0, 0.0),
|
||||
1.0);
|
||||
}
|
||||
} else {
|
||||
meta.size[0] = 0.0;
|
||||
meta.size[1] = 0.0;
|
||||
meta.size[2] = 0.0;
|
||||
}
|
||||
meta.divergence = static_cast<double>(render.divergence_q) / 32767.0;
|
||||
meta.snap = render.snap;
|
||||
}
|
||||
}
|
||||
}
|
||||
double x = 0.0;
|
||||
double y = 0.0;
|
||||
double z = 0.0;
|
||||
geometry::q_to_adm_xyz(state_.q(object, 0), state_.q(object, 1), state_.q(object, 2), &x,
|
||||
&y, &z);
|
||||
const int slot = object - 1;
|
||||
if (!has_previous_[slot] || previous_[slot][0] != x || previous_[slot][1] != y ||
|
||||
previous_[slot][2] != z) {
|
||||
events_[slot].push_back(OamdEvent{event_sample, x, y, z, ramp_samples});
|
||||
const ObjectMetadata& meta = meta_[slot];
|
||||
const bool position_changed =
|
||||
!has_previous_[slot] || previous_[slot][0] != x || previous_[slot][1] != y ||
|
||||
previous_[slot][2] != z;
|
||||
const bool meta_changed = !has_previous_[slot] || !same_meta(previous_meta_[slot], meta);
|
||||
if (position_changed || meta_changed) {
|
||||
events_[slot].push_back(OamdEvent{event_sample, x, y, z, ramp_samples, meta});
|
||||
previous_[slot][0] = x;
|
||||
previous_[slot][1] = y;
|
||||
previous_[slot][2] = z;
|
||||
previous_meta_[slot] = meta;
|
||||
has_previous_[slot] = true;
|
||||
}
|
||||
}
|
||||
|
||||
@@ -19,6 +19,7 @@ struct OamdEvent {
|
||||
double y = 0.0;
|
||||
double z = 0.0;
|
||||
std::int64_t ramp_samples = 0;
|
||||
ObjectMetadata meta;
|
||||
};
|
||||
|
||||
enum class TrajectoryMode { Compact, Dense64 };
|
||||
@@ -62,6 +63,8 @@ private:
|
||||
std::vector<OamdEvent> events_[kObjectCount];
|
||||
bool has_previous_[kObjectCount] = {};
|
||||
double previous_[kObjectCount][3] = {};
|
||||
ObjectMetadata meta_[kObjectCount];
|
||||
ObjectMetadata previous_meta_[kObjectCount];
|
||||
bool fixed_[kObjectCount] = {};
|
||||
double fixed_xyz_[kObjectCount][3] = {};
|
||||
};
|
||||
|
||||
@@ -1,3 +1,4 @@
|
||||
#include <cstddef>
|
||||
#include <cstring>
|
||||
#include <string>
|
||||
|
||||
@@ -45,6 +46,7 @@ joc::stream::Config to_config(const joc_stream_config& config) {
|
||||
if (config.rosella_filterbank != 0u) {
|
||||
out.rosella_filterbank = config.rosella_filterbank;
|
||||
}
|
||||
if (config.pcm_layout != nullptr) { out.pcm_layout = config.pcm_layout; }
|
||||
return out;
|
||||
}
|
||||
|
||||
@@ -94,6 +96,19 @@ joc_error JOC_CALL joc_stream_push(joc_stream* stream, const joc_stream_buffer*
|
||||
}
|
||||
return pushed;
|
||||
}
|
||||
if (input->kind == JOC_STREAM_IN_PCM_LAYOUT) {
|
||||
if (input->sample_rate != 0u && input->sample_rate != 48000u) {
|
||||
return joc::Status::fail(JOC_ERR_INVALID_ARGUMENT, joc::stage::kRender,
|
||||
"PCM layout input must be 48 kHz");
|
||||
}
|
||||
std::size_t consumed = 0;
|
||||
const joc::Status pushed = stream->instance.push_pcm_layout(
|
||||
input->pcm, input->sample_count, input->channels, &consumed);
|
||||
if (consumed_samples != nullptr) {
|
||||
*consumed_samples = static_cast<std::uint32_t>(consumed);
|
||||
}
|
||||
return pushed;
|
||||
}
|
||||
std::size_t consumed = 0;
|
||||
const std::uint32_t channels = input->channels != 0u ? input->channels : 6u;
|
||||
const joc::Status pushed =
|
||||
@@ -140,8 +155,13 @@ joc_error JOC_CALL joc_stream_reset(joc_stream* stream) {
|
||||
}
|
||||
|
||||
joc_error JOC_CALL joc_stream_status(const joc_stream* stream, joc_stream_status_info* out) {
|
||||
// The info struct only grows by appending fields, so a caller built against
|
||||
// the previous size still gets every field it knows about.
|
||||
constexpr uint32_t kSizeBeforeTruncatedFrames = static_cast<uint32_t>(
|
||||
offsetof(joc_stream_status_info, truncated_joc_frames));
|
||||
if (stream == nullptr || out == nullptr ||
|
||||
out->struct_size != sizeof(joc_stream_status_info)) {
|
||||
(out->struct_size != sizeof(joc_stream_status_info) &&
|
||||
out->struct_size != kSizeBeforeTruncatedFrames)) {
|
||||
return JOC_ERR_INVALID_ARGUMENT;
|
||||
}
|
||||
const joc::stream::Info& info = stream->instance.info();
|
||||
@@ -160,6 +180,9 @@ joc_error JOC_CALL joc_stream_status(const joc_stream* stream, joc_stream_status
|
||||
out->bed_mask_mismatch_frames = info.bed_mask_mismatch_frames;
|
||||
out->output_channels = info.output_channels;
|
||||
out->ended = info.ended;
|
||||
if (out->struct_size > kSizeBeforeTruncatedFrames) {
|
||||
out->truncated_joc_frames = info.truncated_joc_frames;
|
||||
}
|
||||
return JOC_OK;
|
||||
}
|
||||
|
||||
|
||||
@@ -98,7 +98,8 @@ Status JochrtfRuntime::open(const hrtf::Field& field, const hrtf::Kernels& kerne
|
||||
Status JochrtfRuntime::process_block() {
|
||||
double positions[timeline::kTimelineObjects][3] = {};
|
||||
const Status queried = timeline_.positions_at(static_cast<std::int64_t>(processed_samples_),
|
||||
positions);
|
||||
&positions[0][0],
|
||||
timeline::kTimelineObjects);
|
||||
if (!queried.ok()) {
|
||||
return queried;
|
||||
}
|
||||
|
||||
@@ -0,0 +1,106 @@
|
||||
#include "binaural/multichannel_layout.h"
|
||||
|
||||
#include <cstring>
|
||||
|
||||
namespace joc::multichannel {
|
||||
|
||||
namespace {
|
||||
|
||||
constexpr ChannelSpec kStereo[] = {
|
||||
{"L", false, {-1.0, 1.0, 0.0}},
|
||||
{"R", false, {1.0, 1.0, 0.0}},
|
||||
};
|
||||
|
||||
constexpr ChannelSpec kSurround51[] = {
|
||||
{"L", false, {-1.0, 1.0, 0.0}},
|
||||
{"R", false, {1.0, 1.0, 0.0}},
|
||||
{"C", false, {0.0, 1.0, 0.0}},
|
||||
{"LFE", true, {0.0, 1.0, 0.0}},
|
||||
{"BL", false, {-1.0, -1.0, 0.0}},
|
||||
{"BR", false, {1.0, -1.0, 0.0}},
|
||||
};
|
||||
|
||||
constexpr ChannelSpec kSurround71[] = {
|
||||
{"L", false, {-1.0, 1.0, 0.0}},
|
||||
{"R", false, {1.0, 1.0, 0.0}},
|
||||
{"C", false, {0.0, 1.0, 0.0}},
|
||||
{"LFE", true, {0.0, 1.0, 0.0}},
|
||||
{"BL", false, {-1.0, -1.0, 0.0}},
|
||||
{"BR", false, {1.0, -1.0, 0.0}},
|
||||
{"SL", false, {-1.0, 0.0, 0.0}},
|
||||
{"SR", false, {1.0, 0.0, 0.0}},
|
||||
};
|
||||
|
||||
constexpr Layout kLayouts[] = {
|
||||
{"2.0", 2, kStereo},
|
||||
{"5.1", 6, kSurround51},
|
||||
{"7.1", 8, kSurround71},
|
||||
};
|
||||
|
||||
} // namespace
|
||||
|
||||
const Layout* find_layout(const std::string& name) {
|
||||
for (const Layout& layout : kLayouts) {
|
||||
if (name == layout.name) {
|
||||
return &layout;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
const Layout* detect_layout(int channels) {
|
||||
for (const Layout& layout : kLayouts) {
|
||||
if (layout.channels == channels) {
|
||||
return &layout;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
Status object_positions(const Layout& layout, double positions[15][3]) {
|
||||
if (positions == nullptr) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kFoundation,
|
||||
"object position output is null");
|
||||
}
|
||||
std::memset(positions, 0, sizeof(double) * 15 * 3);
|
||||
int slot = 0;
|
||||
for (int channel = 0; channel < layout.channels; ++channel) {
|
||||
const ChannelSpec& spec = layout.specs[channel];
|
||||
if (spec.lfe) {
|
||||
continue;
|
||||
}
|
||||
if (slot >= 15) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kFoundation,
|
||||
"layout needs more than 15 object channels");
|
||||
}
|
||||
for (int axis = 0; axis < 3; ++axis) {
|
||||
positions[slot][axis] = spec.position[axis];
|
||||
}
|
||||
++slot;
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
Status pcm_to_objects16(const float* interleaved, std::size_t frames, const Layout& layout,
|
||||
float* objects16_planar) {
|
||||
if (interleaved == nullptr || objects16_planar == nullptr) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kFoundation,
|
||||
"multichannel PCM buffers are null");
|
||||
}
|
||||
std::memset(objects16_planar, 0, sizeof(float) * frames * 16u);
|
||||
int slot = 0;
|
||||
for (int channel = 0; channel < layout.channels; ++channel) {
|
||||
const ChannelSpec& spec = layout.specs[channel];
|
||||
const int destination = spec.lfe ? 0 : 1 + slot;
|
||||
for (std::size_t frame = 0; frame < frames; ++frame) {
|
||||
objects16_planar[static_cast<std::size_t>(destination) * frames + frame] =
|
||||
interleaved[frame * static_cast<std::size_t>(layout.channels) + channel];
|
||||
}
|
||||
if (!spec.lfe) {
|
||||
++slot;
|
||||
}
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
} // namespace joc::multichannel
|
||||
@@ -0,0 +1,39 @@
|
||||
// Channel order and fixed ADM directions for the multichannel binaural path.
|
||||
#pragma once
|
||||
|
||||
#include <cstddef>
|
||||
#include <cstdint>
|
||||
#include <string>
|
||||
|
||||
#include "foundation/status.h"
|
||||
|
||||
namespace joc::multichannel {
|
||||
|
||||
struct ChannelSpec {
|
||||
const char* label;
|
||||
bool lfe;
|
||||
double position[3];
|
||||
};
|
||||
|
||||
struct Layout {
|
||||
const char* name;
|
||||
int channels;
|
||||
const ChannelSpec* specs;
|
||||
};
|
||||
|
||||
// WAV channel order follows the ffmpeg definitions:
|
||||
// 2.0 = FL FR
|
||||
// 5.1 = FL FR FC LFE BL BR
|
||||
// 7.1 = FL FR FC LFE BL BR SL SR
|
||||
// BL/BR are the rear pair, SL/SR the side pair.
|
||||
const Layout* find_layout(const std::string& name);
|
||||
const Layout* detect_layout(int channels);
|
||||
|
||||
Status object_positions(const Layout& layout, double positions[15][3]);
|
||||
|
||||
// Maps interleaved layout PCM to the JOC source layout: channel 0 is LFE,
|
||||
// channels 1..15 are objects. Unused object channels stay silent.
|
||||
Status pcm_to_objects16(const float* interleaved, std::size_t frames, const Layout& layout,
|
||||
float* objects16_planar);
|
||||
|
||||
} // namespace joc::multichannel
|
||||
@@ -0,0 +1,101 @@
|
||||
#include "binaural/pcm_resampler.h"
|
||||
|
||||
#include <algorithm>
|
||||
#include <cmath>
|
||||
|
||||
namespace joc::multichannel {
|
||||
|
||||
namespace {
|
||||
|
||||
constexpr double kPi = 3.14159265358979323846264338327950288;
|
||||
constexpr int kHalfTaps = 32;
|
||||
|
||||
double bessel_i0(double value) {
|
||||
double sum = 1.0;
|
||||
double term = 1.0;
|
||||
const double half = value * 0.5;
|
||||
for (int k = 1; k < 32; ++k) {
|
||||
term *= (half / k) * (half / k);
|
||||
sum += term;
|
||||
if (term < sum * 1.0e-12) {
|
||||
break;
|
||||
}
|
||||
}
|
||||
return sum;
|
||||
}
|
||||
|
||||
double kaiser_window(double x) {
|
||||
constexpr double kBeta = 8.6;
|
||||
const double ratio = std::max(0.0, 1.0 - (x / kHalfTaps) * (x / kHalfTaps));
|
||||
return bessel_i0(kBeta * std::sqrt(ratio)) / bessel_i0(kBeta);
|
||||
}
|
||||
|
||||
double sinc(double x) {
|
||||
if (std::fabs(x) < 1.0e-12) {
|
||||
return 1.0;
|
||||
}
|
||||
return std::sin(kPi * x) / (kPi * x);
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
Status resample_interleaved(const float* input, std::size_t frames, int channels,
|
||||
std::uint32_t source_rate, std::uint32_t target_rate,
|
||||
std::vector<float>* output) {
|
||||
if (output == nullptr || channels <= 0 || frames == 0) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kFoundation,
|
||||
"resampler needs samples, channels and an output vector");
|
||||
}
|
||||
if (source_rate == 0 || target_rate == 0) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kFoundation,
|
||||
"resampler sample rates must be positive");
|
||||
}
|
||||
if (source_rate == target_rate) {
|
||||
output->assign(input, input + frames * static_cast<std::size_t>(channels));
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
const double step = static_cast<double>(source_rate) / static_cast<double>(target_rate);
|
||||
const double cutoff = (target_rate < source_rate)
|
||||
? static_cast<double>(target_rate) / source_rate
|
||||
: 1.0;
|
||||
const std::size_t output_frames = static_cast<std::size_t>(std::llround(
|
||||
static_cast<double>(frames) * target_rate / source_rate));
|
||||
output->assign(output_frames * static_cast<std::size_t>(channels), 0.0f);
|
||||
|
||||
for (std::size_t out_frame = 0; out_frame < output_frames; ++out_frame) {
|
||||
const double position = static_cast<double>(out_frame) * step;
|
||||
const std::int64_t base = static_cast<std::int64_t>(std::floor(position));
|
||||
const double fraction = position - static_cast<double>(base);
|
||||
double weights[2 * kHalfTaps + 1];
|
||||
double weight_sum = 0.0;
|
||||
for (int tap = -kHalfTaps + 1; tap <= kHalfTaps; ++tap) {
|
||||
const double distance = tap - fraction;
|
||||
const double weight = cutoff * sinc(cutoff * distance) * kaiser_window(distance);
|
||||
weights[tap + kHalfTaps] = weight;
|
||||
weight_sum += weight;
|
||||
}
|
||||
if (std::fabs(weight_sum) < 1.0e-12) {
|
||||
continue;
|
||||
}
|
||||
for (int channel = 0; channel < channels; ++channel) {
|
||||
double sum = 0.0;
|
||||
for (int tap = -kHalfTaps + 1; tap <= kHalfTaps; ++tap) {
|
||||
const std::int64_t index = base + tap;
|
||||
if (index < 0 || index >= static_cast<std::int64_t>(frames)) {
|
||||
continue;
|
||||
}
|
||||
sum += static_cast<double>(
|
||||
input[static_cast<std::size_t>(index) *
|
||||
static_cast<std::size_t>(channels) +
|
||||
channel]) *
|
||||
weights[tap + kHalfTaps];
|
||||
}
|
||||
output->at(out_frame * static_cast<std::size_t>(channels) + channel) =
|
||||
static_cast<float>(sum / weight_sum);
|
||||
}
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
} // namespace joc::multichannel
|
||||
@@ -0,0 +1,18 @@
|
||||
// Rational sample-rate conversion for the multichannel binaural frontend.
|
||||
#pragma once
|
||||
|
||||
#include <cstddef>
|
||||
#include <cstdint>
|
||||
#include <vector>
|
||||
|
||||
#include "foundation/status.h"
|
||||
|
||||
namespace joc::multichannel {
|
||||
|
||||
// Resamples interleaved PCM with a windowed-sinc low-pass. The binaural
|
||||
// runtime is native 48 kHz, so any other input rate passes through this first.
|
||||
Status resample_interleaved(const float* input, std::size_t frames, int channels,
|
||||
std::uint32_t source_rate, std::uint32_t target_rate,
|
||||
std::vector<float>* output);
|
||||
|
||||
} // namespace joc::multichannel
|
||||
+2
-2
@@ -83,7 +83,7 @@ struct Options {
|
||||
};
|
||||
|
||||
const char* kLayoutChoices =
|
||||
"2.0 3.1 4.0 5.1 7.1 5.1.2 5.1.4 7.1.2 7.1.4 9.1.4 9.1.6";
|
||||
"2.0 3.1 4.0 5.1 7.1 5.1.2 5.1.4 7.1.2 7.1.4 9.1.4 9.1.6 6.0";
|
||||
|
||||
void print_usage() {
|
||||
std::printf(
|
||||
@@ -97,7 +97,7 @@ void print_usage() {
|
||||
" input 输入 .m4a/.eac3/.ec3\n"
|
||||
"\n"
|
||||
"模式(默认输出 ADM BWF):\n"
|
||||
" --speaker-layout L 直接扬声器渲染布局,例如 2.0、5.1、7.1.2\n"
|
||||
" --speaker-layout L 直接扬声器渲染布局,例如 2.0、5.1、6.0、7.1.2\n"
|
||||
" 可选值: %s\n"
|
||||
" --binaural 直接双耳渲染;不生成临时 ADM BWF\n"
|
||||
"\n"
|
||||
|
||||
@@ -0,0 +1,296 @@
|
||||
// Multichannel PCM WAV -> binaural WAV frontend.
|
||||
//
|
||||
// Ordinary 2.0/5.1/7.1 PCM has no OAMD. Each non-LFE channel is assigned a
|
||||
// fixed ADM direction, the LFE channel goes to source 0, and the resulting
|
||||
// sixteen JOC sources feed the Rosella binaural runtime. The runtime is
|
||||
// native 48 kHz; other input rates are resampled on the way in and the stereo
|
||||
// result is resampled to the requested output rate on the way out.
|
||||
#include <algorithm>
|
||||
#include <cmath>
|
||||
#include <cstdint>
|
||||
#include <cstdio>
|
||||
#include <cstdlib>
|
||||
#include <cstring>
|
||||
#include <string>
|
||||
#include <vector>
|
||||
|
||||
#include "binaural/multichannel_layout.h"
|
||||
#include "binaural/pcm_resampler.h"
|
||||
#include "foundation/status.h"
|
||||
#include "hrtf/rosella_model.h"
|
||||
#include "hrtf/rosella_renderer.h"
|
||||
#include "io/wav_pcm.h"
|
||||
#include "oamd/oamd_parser.h"
|
||||
|
||||
namespace {
|
||||
|
||||
constexpr std::uint32_t kFrameSamples = 1536;
|
||||
constexpr std::uint32_t kSourceChannels = 16;
|
||||
constexpr std::uint32_t kTargetRate = 48000;
|
||||
|
||||
void usage() {
|
||||
std::printf(
|
||||
"usage: joc_multichannel INPUT.wav -o OUTPUT.wav [options]\n"
|
||||
" --layout auto|2.0|5.1|7.1 WAV channel layout (auto by channel count)\n"
|
||||
" --model PATH Rosella .personalized_headphone model\n"
|
||||
" --mode near|mid|far Rosella binaural profile (default mid)\n"
|
||||
" --filterbank 1|2 Rosella filterbank (default 1)\n"
|
||||
" --tail-seconds SECONDS room tail (default 2.0)\n"
|
||||
" --output-rate 48000|input|N output rate (default 48000; non-48k adds\n"
|
||||
" a second resampling pass)\n"
|
||||
" --list-layouts\n"
|
||||
"\n"
|
||||
"Prefer a 48 kHz float32 WAV; let ffmpeg/libsoxr do the one conversion:\n"
|
||||
" ffmpeg -i INPUT -af \"aresample=resampler=soxr:precision=28:osr=48000\" "
|
||||
"-c:a pcm_f32le OUT.wav\n"
|
||||
"44.1/88.2/176.4 or 48/96/192 kHz accepted; non-48k is converted once in.\n");
|
||||
}
|
||||
|
||||
std::int16_t q_for_position(double value) {
|
||||
if (value <= -0.999) {
|
||||
return 0;
|
||||
}
|
||||
if (value >= 0.999) {
|
||||
return 32767;
|
||||
}
|
||||
return 16384;
|
||||
}
|
||||
|
||||
void fill_initial_update(const joc::multichannel::Layout& layout, joc::oamd::OamdUpdate* update) {
|
||||
update->has_frame_state = true;
|
||||
update->object_count = kSourceChannels;
|
||||
update->frame_state.object_count = kSourceChannels;
|
||||
update->frame_state.block_count = 1;
|
||||
update->frame_state.blocks.resize(1);
|
||||
update->frame_state.objects.resize(kSourceChannels);
|
||||
double positions[15][3];
|
||||
const joc::Status status = joc::multichannel::object_positions(layout, positions);
|
||||
if (!status.ok()) {
|
||||
return;
|
||||
}
|
||||
for (int object = 0; object < static_cast<int>(kSourceChannels); ++object) {
|
||||
joc::oamd::ObjectElementState& element = update->frame_state.objects[object];
|
||||
element.kind = object == 0 ? joc::oamd::ObjectKind::kBed
|
||||
: joc::oamd::ObjectKind::kDynamic;
|
||||
element.blocks.resize(1);
|
||||
joc::oamd::ObjectRenderState& render = element.blocks[0].render;
|
||||
render.not_active = false;
|
||||
render.gain_q = 4096;
|
||||
render.priority_q = 32767;
|
||||
render.enable_elevation = true;
|
||||
double x = 0.0;
|
||||
double y = 1.0;
|
||||
double z = 0.0;
|
||||
if (object > 0) {
|
||||
x = positions[object - 1][0];
|
||||
y = positions[object - 1][1];
|
||||
z = positions[object - 1][2];
|
||||
}
|
||||
render.qx = q_for_position(x);
|
||||
render.qy = q_for_position(-y);
|
||||
render.qz = static_cast<std::int32_t>(std::llround(z * 32767.0));
|
||||
}
|
||||
}
|
||||
|
||||
std::string model_default() {
|
||||
return "HRTF/binaural.personalized_headphone";
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
int main(int argc, char** argv) {
|
||||
std::string input;
|
||||
std::string output;
|
||||
std::string layout_name = "auto";
|
||||
std::string model_path = model_default();
|
||||
std::string mode = "mid";
|
||||
int filterbank = 1;
|
||||
double tail_seconds = 2.0;
|
||||
std::string output_rate = "48000";
|
||||
|
||||
for (int i = 1; i < argc; ++i) {
|
||||
const std::string arg = argv[i];
|
||||
const auto next = [&](const char* option) -> std::string {
|
||||
if (i + 1 >= argc) {
|
||||
std::fprintf(stderr, "missing value for %s\n", option);
|
||||
std::exit(2);
|
||||
}
|
||||
return argv[++i];
|
||||
};
|
||||
if (arg == "--layout") {
|
||||
layout_name = next("--layout");
|
||||
} else if (arg == "--model") {
|
||||
model_path = next("--model");
|
||||
} else if (arg == "--mode") {
|
||||
mode = next("--mode");
|
||||
} else if (arg == "--filterbank") {
|
||||
filterbank = std::stoi(next("--filterbank"));
|
||||
} else if (arg == "--tail-seconds") {
|
||||
tail_seconds = std::stod(next("--tail-seconds"));
|
||||
} else if (arg == "--output-rate") {
|
||||
output_rate = next("--output-rate");
|
||||
} else if (arg == "--list-layouts") {
|
||||
std::printf("2.0: 2 channels\n5.1: 6 channels\n7.1: 8 channels\n");
|
||||
return 0;
|
||||
} else if (arg == "-o" || arg == "--output") {
|
||||
output = next("-o");
|
||||
} else if (arg == "-h" || arg == "--help") {
|
||||
usage();
|
||||
return 0;
|
||||
} else if (!arg.empty() && arg[0] == '-') {
|
||||
std::fprintf(stderr, "unknown option: %s\n", arg.c_str());
|
||||
return 2;
|
||||
} else if (input.empty()) {
|
||||
input = arg;
|
||||
} else {
|
||||
std::fprintf(stderr, "unexpected argument: %s\n", arg.c_str());
|
||||
return 2;
|
||||
}
|
||||
}
|
||||
if (input.empty()) {
|
||||
usage();
|
||||
return 2;
|
||||
}
|
||||
if (output.empty()) {
|
||||
output = input + ".binaural.wav";
|
||||
}
|
||||
|
||||
joc::io::WavAudio wav;
|
||||
joc::Status status = joc::io::read_wav_file(input, &wav);
|
||||
if (!status.ok()) {
|
||||
std::fprintf(stderr, "read: %s\n", status.message().c_str());
|
||||
return 1;
|
||||
}
|
||||
const joc::multichannel::Layout* layout =
|
||||
(layout_name == "auto") ? joc::multichannel::detect_layout(wav.channels)
|
||||
: joc::multichannel::find_layout(layout_name);
|
||||
if (layout == nullptr) {
|
||||
std::fprintf(stderr, "unsupported layout: %s (%u channels)\n", layout_name.c_str(),
|
||||
static_cast<unsigned>(wav.channels));
|
||||
return 2;
|
||||
}
|
||||
if (wav.channels != layout->channels) {
|
||||
std::fprintf(stderr, "layout %s expects %d channels, WAV has %u\n", layout->name,
|
||||
layout->channels, static_cast<unsigned>(wav.channels));
|
||||
return 2;
|
||||
}
|
||||
|
||||
std::vector<float> pcm48;
|
||||
if (wav.sample_rate == kTargetRate) {
|
||||
pcm48 = wav.samples;
|
||||
} else {
|
||||
std::fprintf(
|
||||
stderr,
|
||||
"warning: input is %u Hz and was converted once to 48000 Hz; prefer "
|
||||
"ffmpeg/libsoxr to produce 48 kHz float32 first:\n"
|
||||
" ffmpeg -i INPUT -af \"aresample=resampler=soxr:precision=28:osr=48000\" "
|
||||
"-c:a pcm_f32le OUT.wav\n",
|
||||
static_cast<unsigned>(wav.sample_rate));
|
||||
status = joc::multichannel::resample_interleaved(
|
||||
wav.samples.data(), wav.samples.size() / wav.channels, wav.channels,
|
||||
wav.sample_rate, kTargetRate, &pcm48);
|
||||
if (!status.ok()) {
|
||||
std::fprintf(stderr, "resample: %s\n", status.message().c_str());
|
||||
return 1;
|
||||
}
|
||||
}
|
||||
const std::size_t frames = pcm48.size() / wav.channels;
|
||||
|
||||
joc::hrtf::RosellaModel model;
|
||||
status = joc::hrtf::load_personalized_headphone(model_path, &model);
|
||||
if (!status.ok()) {
|
||||
std::fprintf(stderr, "model: %s\n", status.message().c_str());
|
||||
return 1;
|
||||
}
|
||||
joc::hrtf::RosellaRenderOptions options;
|
||||
options.profile = (mode == "near") ? joc::hrtf::RosellaProfile::Near
|
||||
: (mode == "far") ? joc::hrtf::RosellaProfile::Far
|
||||
: joc::hrtf::RosellaProfile::Mid;
|
||||
options.bank = (filterbank == 2) ? joc::hrtf::RosellaBankKind::Captured
|
||||
: joc::hrtf::RosellaBankKind::Public;
|
||||
options.tail_seconds = tail_seconds;
|
||||
joc::hrtf::RosellaRuntime runtime;
|
||||
status = runtime.open(model, options);
|
||||
if (!status.ok()) {
|
||||
std::fprintf(stderr, "open: %s\n", status.message().c_str());
|
||||
return 1;
|
||||
}
|
||||
|
||||
joc::oamd::OamdUpdate initial{};
|
||||
fill_initial_update(*layout, &initial);
|
||||
std::vector<float> objects16(kSourceChannels * kFrameSamples, 0.0f);
|
||||
std::vector<double> rendered;
|
||||
for (std::size_t start = 0, frame = 0; start < frames; start += kFrameSamples, ++frame) {
|
||||
const std::size_t count = std::min<std::size_t>(kFrameSamples, frames - start);
|
||||
std::fill(objects16.begin(), objects16.end(), 0.0f);
|
||||
for (int channel = 0; channel < layout->channels; ++channel) {
|
||||
const joc::multichannel::ChannelSpec& spec = layout->specs[channel];
|
||||
int destination = 0;
|
||||
if (!spec.lfe) {
|
||||
destination = 1;
|
||||
for (int previous = 0; previous < channel; ++previous) {
|
||||
if (!layout->specs[previous].lfe) {
|
||||
++destination;
|
||||
}
|
||||
}
|
||||
}
|
||||
for (std::size_t sample = 0; sample < count; ++sample) {
|
||||
objects16[static_cast<std::size_t>(destination) * kFrameSamples + sample] =
|
||||
pcm48[(start + sample) * layout->channels + channel];
|
||||
}
|
||||
}
|
||||
status = runtime.submit_frame(objects16.data(), frame == 0 ? &initial : nullptr,
|
||||
static_cast<std::int64_t>(frame), 0, 0);
|
||||
if (!status.ok()) {
|
||||
std::fprintf(stderr, "render: %s\n", status.message().c_str());
|
||||
return 1;
|
||||
}
|
||||
std::vector<double> chunk;
|
||||
runtime.take_output(&chunk);
|
||||
rendered.insert(rendered.end(), chunk.begin(), chunk.end());
|
||||
}
|
||||
std::vector<double> tail;
|
||||
status = runtime.finish(runtime.finish_capacity(tail_seconds), &tail);
|
||||
if (!status.ok()) {
|
||||
std::fprintf(stderr, "finish: %s\n", status.message().c_str());
|
||||
return 1;
|
||||
}
|
||||
rendered.insert(rendered.end(), tail.begin(), tail.end());
|
||||
|
||||
std::vector<float> stereo(rendered.size());
|
||||
for (std::size_t index = 0; index < rendered.size(); ++index) {
|
||||
stereo[index] = static_cast<float>(rendered[index]);
|
||||
}
|
||||
std::uint32_t target_rate = wav.sample_rate;
|
||||
if (output_rate == "48k" || output_rate == "48000") {
|
||||
target_rate = kTargetRate;
|
||||
} else if (output_rate != "input") {
|
||||
target_rate = static_cast<std::uint32_t>(std::stoul(output_rate));
|
||||
}
|
||||
if (target_rate != kTargetRate) {
|
||||
std::fprintf(stderr,
|
||||
"warning: non-48k output adds a second resampling pass; the "
|
||||
"native binaural output rate is 48000 Hz\n");
|
||||
std::vector<float> converted;
|
||||
status = joc::multichannel::resample_interleaved(
|
||||
stereo.data(), stereo.size() / 2, 2, kTargetRate, target_rate, &converted);
|
||||
if (!status.ok()) {
|
||||
std::fprintf(stderr, "output resample: %s\n", status.message().c_str());
|
||||
return 1;
|
||||
}
|
||||
stereo.swap(converted);
|
||||
}
|
||||
status = joc::io::write_wav_float32(output, target_rate, 2, stereo.data(),
|
||||
stereo.size() / 2);
|
||||
if (!status.ok()) {
|
||||
std::fprintf(stderr, "write: %s\n", status.message().c_str());
|
||||
return 1;
|
||||
}
|
||||
std::printf("input: %s (%u ch, %u Hz, %zu frames)\n", input.c_str(),
|
||||
static_cast<unsigned>(wav.channels), static_cast<unsigned>(wav.sample_rate),
|
||||
frames);
|
||||
std::printf("layout: %s\n", layout->name);
|
||||
std::printf("output: %s (%u Hz, %zu frames)\n", output.c_str(),
|
||||
static_cast<unsigned>(target_rate), stereo.size() / 2);
|
||||
return 0;
|
||||
}
|
||||
@@ -102,9 +102,15 @@ Status parse_at(const std::uint8_t* data, std::size_t size, std::size_t start_bi
|
||||
}
|
||||
|
||||
const std::uint32_t has_sample_offset = reader.read(1);
|
||||
std::uint16_t sample_offset = 0;
|
||||
std::uint32_t sample_offset = 0;
|
||||
if (has_sample_offset != 0u) {
|
||||
sample_offset = static_cast<std::uint16_t>(reader.read(12) >> 1);
|
||||
std::uint32_t value = 0;
|
||||
if (!bits::variable_bits(reader, 11, 8, &value)) {
|
||||
return reader.error() == JOC_ERR_BITSTREAM_TRUNCATED
|
||||
? truncated_fail(reader)
|
||||
: syntax_fail(reader.error_message());
|
||||
}
|
||||
sample_offset = value;
|
||||
}
|
||||
if (reader.read(1) != 0u) {
|
||||
std::uint32_t ignored = 0;
|
||||
|
||||
@@ -17,7 +17,7 @@ inline constexpr std::size_t kMaxPayloads = JOC_MAX_EMDF_PAYLOADS;
|
||||
|
||||
struct Payload {
|
||||
std::uint8_t id = 0;
|
||||
std::uint16_t sample_offset = 0;
|
||||
std::uint32_t sample_offset = 0; // EMDF outer smpoffst: variable_bits(11)
|
||||
std::size_t bit_offset = 0; // MSB-first bit position of the payload bytes
|
||||
std::size_t size = 0; // payload byte count
|
||||
// Payload configuration data (ETSI TS 102 366 H.2.2.3). Spec section 8.2 and
|
||||
|
||||
@@ -1120,7 +1120,8 @@ struct RosellaRuntime::Impl {
|
||||
// rosella_binaural_renderer._set_block_parameters.
|
||||
Status set_block_parameters(std::int64_t sample) {
|
||||
double positions[timeline::kTimelineObjects][3] = {};
|
||||
Status status = timeline.positions_at(sample, positions);
|
||||
Status status = timeline.positions_at(sample, &positions[0][0],
|
||||
timeline::kTimelineObjects);
|
||||
if (!status.ok()) {
|
||||
return status;
|
||||
}
|
||||
|
||||
@@ -0,0 +1,207 @@
|
||||
#include "io/wav_pcm.h"
|
||||
|
||||
#include <cstring>
|
||||
#include <fstream>
|
||||
#include <iterator>
|
||||
#include <limits>
|
||||
|
||||
namespace joc::io {
|
||||
|
||||
namespace {
|
||||
|
||||
std::uint16_t read_u16(const std::uint8_t* data) {
|
||||
return static_cast<std::uint16_t>(data[0] | (data[1] << 8));
|
||||
}
|
||||
|
||||
std::uint32_t read_u32(const std::uint8_t* data) {
|
||||
return static_cast<std::uint32_t>(data[0]) |
|
||||
(static_cast<std::uint32_t>(data[1]) << 8) |
|
||||
(static_cast<std::uint32_t>(data[2]) << 16) |
|
||||
(static_cast<std::uint32_t>(data[3]) << 24);
|
||||
}
|
||||
|
||||
void write_u16(std::vector<std::uint8_t>* out, std::uint16_t value) {
|
||||
out->push_back(static_cast<std::uint8_t>(value & 0xFFu));
|
||||
out->push_back(static_cast<std::uint8_t>((value >> 8) & 0xFFu));
|
||||
}
|
||||
|
||||
void write_u32(std::vector<std::uint8_t>* out, std::uint32_t value) {
|
||||
out->push_back(static_cast<std::uint8_t>(value & 0xFFu));
|
||||
out->push_back(static_cast<std::uint8_t>((value >> 8) & 0xFFu));
|
||||
out->push_back(static_cast<std::uint8_t>((value >> 16) & 0xFFu));
|
||||
out->push_back(static_cast<std::uint8_t>((value >> 24) & 0xFFu));
|
||||
}
|
||||
|
||||
std::string fourcc(const std::uint8_t* data) {
|
||||
return std::string(reinterpret_cast<const char*>(data), 4);
|
||||
}
|
||||
|
||||
float sample_to_float(const std::uint8_t* data, std::uint16_t format, std::uint16_t bits) {
|
||||
if (format == 3u) {
|
||||
if (bits == 32u) {
|
||||
float value = 0.0f;
|
||||
std::memcpy(&value, data, sizeof(value));
|
||||
return value;
|
||||
}
|
||||
double value = 0.0;
|
||||
std::memcpy(&value, data, sizeof(value));
|
||||
return static_cast<float>(value);
|
||||
}
|
||||
if (bits == 16u) {
|
||||
const std::int16_t value = static_cast<std::int16_t>(read_u16(data));
|
||||
return static_cast<float>(value) / 32768.0f;
|
||||
}
|
||||
if (bits == 24u) {
|
||||
std::int32_t value = static_cast<std::int32_t>(data[0]) |
|
||||
(static_cast<std::int32_t>(data[1]) << 8) |
|
||||
(static_cast<std::int32_t>(data[2]) << 16);
|
||||
if ((value & 0x00800000) != 0) {
|
||||
value |= static_cast<std::int32_t>(0xFF000000u);
|
||||
}
|
||||
return static_cast<float>(value) / 8388608.0f;
|
||||
}
|
||||
if (bits == 32u) {
|
||||
const std::int32_t value = static_cast<std::int32_t>(read_u32(data));
|
||||
return static_cast<float>(static_cast<double>(value) / 2147483648.0);
|
||||
}
|
||||
return 0.0f;
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
Status read_wav_file(const std::string& path, WavAudio* out) {
|
||||
if (out == nullptr) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kFoundation,
|
||||
"WAV output pointer is null");
|
||||
}
|
||||
std::ifstream stream(path, std::ios::binary);
|
||||
if (!stream) {
|
||||
return Status::fail(JOC_ERR_IO, stage::kFoundation,
|
||||
"cannot open WAV file: " + path);
|
||||
}
|
||||
std::vector<std::uint8_t> bytes((std::istreambuf_iterator<char>(stream)),
|
||||
std::istreambuf_iterator<char>());
|
||||
if (bytes.size() < 12u || fourcc(bytes.data()) != "RIFF" ||
|
||||
fourcc(bytes.data() + 8) != "WAVE") {
|
||||
return Status::fail(JOC_ERR_INPUT_FORMAT, stage::kFoundation,
|
||||
"not a RIFF/WAVE file: " + path);
|
||||
}
|
||||
|
||||
std::uint16_t format = 0;
|
||||
std::uint16_t channels = 0;
|
||||
std::uint32_t sample_rate = 0;
|
||||
std::uint16_t block_align = 0;
|
||||
std::uint16_t bits = 0;
|
||||
bool have_format = false;
|
||||
std::size_t data_offset = 0;
|
||||
std::size_t data_size = 0;
|
||||
bool have_data = false;
|
||||
|
||||
std::size_t offset = 12;
|
||||
while (offset + 8u <= bytes.size()) {
|
||||
const std::string id = fourcc(bytes.data() + offset);
|
||||
const std::uint32_t size = read_u32(bytes.data() + offset + 4);
|
||||
const std::size_t payload = offset + 8u;
|
||||
if (payload + size > bytes.size()) {
|
||||
return Status::fail(JOC_ERR_INPUT_FORMAT, stage::kFoundation,
|
||||
"truncated WAV chunk: " + id);
|
||||
}
|
||||
if (id == "fmt " && size >= 16u) {
|
||||
format = read_u16(bytes.data() + payload);
|
||||
channels = read_u16(bytes.data() + payload + 2);
|
||||
sample_rate = read_u32(bytes.data() + payload + 4);
|
||||
block_align = read_u16(bytes.data() + payload + 12);
|
||||
bits = read_u16(bytes.data() + payload + 14);
|
||||
if (format == 0xFFFEu && size >= 40u) {
|
||||
// WAVEFORMATEXTENSIBLE: the sub-format GUID starts at +24.
|
||||
format = read_u16(bytes.data() + payload + 24);
|
||||
}
|
||||
have_format = true;
|
||||
} else if (id == "data") {
|
||||
data_offset = payload;
|
||||
data_size = size;
|
||||
have_data = true;
|
||||
}
|
||||
offset = payload + size + (size & 1u);
|
||||
}
|
||||
|
||||
if (!have_format || !have_data || channels == 0 || sample_rate == 0) {
|
||||
return Status::fail(JOC_ERR_INPUT_FORMAT, stage::kFoundation,
|
||||
"WAV file has no usable fmt/data chunk: " + path);
|
||||
}
|
||||
if (format != 1u && format != 3u) {
|
||||
return Status::fail(JOC_ERR_INPUT_FORMAT, stage::kFoundation,
|
||||
"unsupported WAV sample format");
|
||||
}
|
||||
const std::uint16_t bytes_per_sample = static_cast<std::uint16_t>(bits / 8u);
|
||||
if (bytes_per_sample == 0u) {
|
||||
return Status::fail(JOC_ERR_INPUT_FORMAT, stage::kFoundation,
|
||||
"unsupported WAV bit depth");
|
||||
}
|
||||
if (block_align == 0u) {
|
||||
block_align = static_cast<std::uint16_t>(channels * bytes_per_sample);
|
||||
}
|
||||
if (block_align < channels * bytes_per_sample) {
|
||||
return Status::fail(JOC_ERR_INPUT_FORMAT, stage::kFoundation,
|
||||
"invalid WAV block alignment");
|
||||
}
|
||||
const std::size_t frames = data_size / block_align;
|
||||
out->sample_rate = sample_rate;
|
||||
out->channels = channels;
|
||||
out->samples.resize(frames * channels);
|
||||
const std::uint8_t* source = bytes.data() + data_offset;
|
||||
for (std::size_t frame = 0; frame < frames; ++frame) {
|
||||
for (std::size_t channel = 0; channel < channels; ++channel) {
|
||||
out->samples[frame * channels + channel] = sample_to_float(
|
||||
source + frame * block_align + channel * bytes_per_sample, format, bits);
|
||||
}
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
Status write_wav_float32(const std::string& path, std::uint32_t sample_rate,
|
||||
std::uint16_t channels, const float* interleaved, std::size_t frames) {
|
||||
if (channels == 0u || sample_rate == 0u || interleaved == nullptr) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kOutput,
|
||||
"WAV writer needs channels, a sample rate and samples");
|
||||
}
|
||||
if (frames > (std::numeric_limits<std::uint32_t>::max() - 36u) /
|
||||
(channels * sizeof(float))) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kOutput,
|
||||
"WAV output is too large for RIFF");
|
||||
}
|
||||
const std::uint32_t data_bytes =
|
||||
static_cast<std::uint32_t>(frames * channels * sizeof(float));
|
||||
std::vector<std::uint8_t> header;
|
||||
header.reserve(44);
|
||||
header.insert(header.end(), {'R', 'I', 'F', 'F'});
|
||||
write_u32(&header, 36u + data_bytes);
|
||||
header.insert(header.end(), {'W', 'A', 'V', 'E'});
|
||||
header.insert(header.end(), {'f', 'm', 't', ' '});
|
||||
write_u32(&header, 16u);
|
||||
write_u16(&header, 3u); // IEEE float
|
||||
write_u16(&header, channels);
|
||||
write_u32(&header, sample_rate);
|
||||
write_u32(&header, sample_rate * static_cast<std::uint32_t>(channels * sizeof(float)));
|
||||
write_u16(&header, static_cast<std::uint16_t>(channels * sizeof(float)));
|
||||
write_u16(&header, 32u);
|
||||
header.insert(header.end(), {'d', 'a', 't', 'a'});
|
||||
write_u32(&header, data_bytes);
|
||||
|
||||
std::ofstream stream(path, std::ios::binary);
|
||||
if (!stream) {
|
||||
return Status::fail(JOC_ERR_IO, stage::kOutput,
|
||||
"cannot open WAV output: " + path);
|
||||
}
|
||||
stream.write(reinterpret_cast<const char*>(header.data()),
|
||||
static_cast<std::streamsize>(header.size()));
|
||||
stream.write(reinterpret_cast<const char*>(interleaved),
|
||||
static_cast<std::streamsize>(data_bytes));
|
||||
if (!stream) {
|
||||
return Status::fail(JOC_ERR_IO, stage::kOutput,
|
||||
"cannot write WAV output: " + path);
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
} // namespace joc::io
|
||||
@@ -0,0 +1,25 @@
|
||||
// Minimal RIFF/WAVE reader and float32 writer for the multichannel binaural
|
||||
// frontend. It accepts the formats ffmpeg writes for PCM input: 16/24/32-bit
|
||||
// integer, 32/64-bit float, including WAVE_FORMAT_EXTENSIBLE.
|
||||
#pragma once
|
||||
|
||||
#include <cstdint>
|
||||
#include <string>
|
||||
#include <vector>
|
||||
|
||||
#include "foundation/status.h"
|
||||
|
||||
namespace joc::io {
|
||||
|
||||
struct WavAudio {
|
||||
std::uint32_t sample_rate = 0;
|
||||
std::uint16_t channels = 0;
|
||||
std::vector<float> samples; // interleaved
|
||||
};
|
||||
|
||||
Status read_wav_file(const std::string& path, WavAudio* out);
|
||||
|
||||
Status write_wav_float32(const std::string& path, std::uint32_t sample_rate,
|
||||
std::uint16_t channels, const float* interleaved, std::size_t frames);
|
||||
|
||||
} // namespace joc::io
|
||||
@@ -1,6 +1,7 @@
|
||||
|
||||
#include "joc_bitstream/bed_layout_map.h"
|
||||
|
||||
#include <cmath>
|
||||
#include <cstring>
|
||||
|
||||
#include "speaker/speaker_layouts.h"
|
||||
@@ -100,6 +101,9 @@ constexpr LayoutName kLayoutNames[] = {
|
||||
{bit(BedChannel::kLeft) | bit(BedChannel::kRight) |
|
||||
bit(BedChannel::kLeftSurround) | bit(BedChannel::kRightSurround) |
|
||||
bit(BedChannel::kCenter) | bit(BedChannel::kLfe), "5.1"},
|
||||
{bit(BedChannel::kLeft) | bit(BedChannel::kRight) |
|
||||
bit(BedChannel::kLeftSurround) | bit(BedChannel::kRightSurround) |
|
||||
bit(BedChannel::kLeftRearSurround) | bit(BedChannel::kRightRearSurround), "6.0"},
|
||||
{bit(BedChannel::kLeft) | bit(BedChannel::kRight) |
|
||||
bit(BedChannel::kLeftSurround) | bit(BedChannel::kRightSurround) |
|
||||
bit(BedChannel::kCenter) | bit(BedChannel::kLfe) |
|
||||
@@ -328,4 +332,301 @@ bool layout_bed_anchor(std::uint32_t out_ch_config, BedChannel channel,
|
||||
return false;
|
||||
}
|
||||
|
||||
|
||||
namespace {
|
||||
|
||||
// ETSI TS 103 420 Table B.10: the fixed ADM position of every Table 12 bed
|
||||
// channel. bed_output_description() matches these points against the target
|
||||
// layout; the kChannels table above keeps the ADM/binaural path's coordinates.
|
||||
struct TableB10Position {
|
||||
BedChannel channel;
|
||||
double xyz[3];
|
||||
};
|
||||
|
||||
constexpr TableB10Position kTableB10[oamd::kBedLabelCount] = {
|
||||
{BedChannel::kLeft, {-1.0, 1.0, 0.0}},
|
||||
{BedChannel::kRight, {1.0, 1.0, 0.0}},
|
||||
{BedChannel::kCenter, {0.0, 1.0, 0.0}},
|
||||
{BedChannel::kLfe, {-1.0, 1.0, -1.0}},
|
||||
{BedChannel::kLeftSurround, {-1.0, 0.0, 0.0}},
|
||||
{BedChannel::kRightSurround, {1.0, 0.0, 0.0}},
|
||||
{BedChannel::kLeftRearSurround, {-1.0, -1.0, 0.0}},
|
||||
{BedChannel::kRightRearSurround, {1.0, -1.0, 0.0}},
|
||||
{BedChannel::kTopFrontLeft, {-1.0, 1.0, 1.0}},
|
||||
{BedChannel::kTopFrontRight, {1.0, 1.0, 1.0}},
|
||||
{BedChannel::kTopSideLeft, {-1.0, 0.0, 1.0}},
|
||||
{BedChannel::kTopSideRight, {1.0, 0.0, 1.0}},
|
||||
{BedChannel::kTopBackLeft, {-1.0, -1.0, 1.0}},
|
||||
{BedChannel::kTopBackRight, {1.0, -1.0, 1.0}},
|
||||
{BedChannel::kLeftWide, {-1.0, 21.0 / 31.0, 0.0}},
|
||||
{BedChannel::kRightWide, {1.0, 21.0 / 31.0, 0.0}},
|
||||
{BedChannel::kLfe2, {1.0, 1.0, -1.0}},
|
||||
};
|
||||
|
||||
// bed_layout.BED_REFERENCE_POSITIONS_Q15; index 3 is the non-spatial sentinel.
|
||||
constexpr std::int32_t kReferencePositions[kBedReferencePositionCount][3] = {
|
||||
{0, 0, 0}, {32767, 0, 0}, {16384, 0, 0}, {-1, -1, -1},
|
||||
{0, 16384, 0}, {32767, 16384, 0}, {0, 32767, 0}, {32767, 32767, 0},
|
||||
{16384, 32767, 0}, {5285, 0, 0}, {27483, 0, 0}, {0, 5285, 0},
|
||||
{32767, 5285, 0}, {7928, 0, 32767}, {24840, 0, 32767}, {7928, 7928, 32767},
|
||||
{24840, 7928, 32767}, {7928, 16384, 32767}, {24840, 16384, 32767},
|
||||
{7928, 24840, 32767}, {24840, 24840, 32767}, {7928, 32767, 32767},
|
||||
{24840, 32767, 32767}, {11099, 0, 0}, {21669, 0, 0}, {0, 11099, 0},
|
||||
{32767, 11099, 0}, {0, 21669, 0}, {32767, 21669, 0}, {0, 27483, 0},
|
||||
{32767, 27483, 0}, {5285, 32767, 0}, {27483, 32767, 0}, {11099, 32767, 0},
|
||||
{21669, 32767, 0}, {16384, 0, 32767}, {16384, 16384, 32767},
|
||||
{16384, 32767, 32767},
|
||||
};
|
||||
|
||||
// 自动 trim 的每个布局基准衰减,单位 dB。
|
||||
struct AttenuationEntry {
|
||||
const char* compact_name;
|
||||
double db;
|
||||
};
|
||||
|
||||
constexpr AttenuationEntry kAttenuation[] = {
|
||||
{"20", -4.5}, {"31", -4.5}, {"40", -3.0}, {"51", -3.0},
|
||||
{"60", -1.5}, {"71", -1.5}, {"512", -2.25}, {"514", -1.5},
|
||||
{"712", -0.75}, {"714", 0.0}, {"914", 0.0}, {"916", 0.0},
|
||||
};
|
||||
|
||||
constexpr int kMaxChannels = static_cast<int>(ejoc::speaker_tables::kMaxChannels);
|
||||
|
||||
void set_sentinel(std::int32_t coordinate[3]) {
|
||||
coordinate[0] = -1;
|
||||
coordinate[1] = -1;
|
||||
coordinate[2] = -1;
|
||||
}
|
||||
|
||||
bool is_sentinel(const std::int32_t coordinate[3]) {
|
||||
return coordinate[0] == -1 && coordinate[1] == -1 && coordinate[2] == -1;
|
||||
}
|
||||
|
||||
bool is_lfe_label(BedChannel channel) {
|
||||
return channel == BedChannel::kLfe || channel == BedChannel::kLfe2;
|
||||
}
|
||||
|
||||
const TableB10Position* table_b10(BedChannel channel) {
|
||||
for (const TableB10Position& entry : kTableB10) {
|
||||
if (entry.channel == channel) {
|
||||
return &entry;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
// ADM 笛卡尔坐标 -> 渲染器方位 (u 右, v 后, w 上)。
|
||||
void render_coordinates(const double xyz[3], double out[3]) {
|
||||
out[0] = (xyz[0] + 1.0) * 0.5;
|
||||
out[1] = (1.0 - xyz[1]) * 0.5;
|
||||
out[2] = xyz[2];
|
||||
}
|
||||
|
||||
// 渲染器方位的欧氏距离,比较的是归一化到 0..1 的坐标。
|
||||
double render_distance(const double left[3], const double right[3]) {
|
||||
const double du = left[0] - right[0];
|
||||
const double dv = left[1] - right[1];
|
||||
const double dw = left[2] - right[2];
|
||||
return std::sqrt(du * du + dv * dv + dw * dw);
|
||||
}
|
||||
|
||||
// Reference-position matching: the non-spatial sentinel resolves to index 3.
|
||||
int nearest_reference(const std::int32_t coordinate[3]) {
|
||||
if (is_sentinel(coordinate)) {
|
||||
return 3;
|
||||
}
|
||||
const double position[3] = {
|
||||
static_cast<double>(coordinate[0]) / 32768.0,
|
||||
static_cast<double>(coordinate[1]) / 32768.0,
|
||||
static_cast<double>(coordinate[2]) / 32768.0,
|
||||
};
|
||||
int best = 3;
|
||||
double best_distance = 0.0;
|
||||
bool have_best = false;
|
||||
for (int index = 0; index < kBedReferencePositionCount; ++index) {
|
||||
const std::int32_t* candidate = kReferencePositions[index];
|
||||
if (candidate[0] == -1 && candidate[1] == -1 && candidate[2] == -1) {
|
||||
continue;
|
||||
}
|
||||
const double point[3] = {
|
||||
static_cast<double>(candidate[0]) / 32768.0,
|
||||
static_cast<double>(candidate[1]) / 32768.0,
|
||||
static_cast<double>(candidate[2]) / 32768.0,
|
||||
};
|
||||
const double distance = render_distance(position, point);
|
||||
if (!have_best || distance < best_distance) {
|
||||
best = index;
|
||||
best_distance = distance;
|
||||
have_best = true;
|
||||
}
|
||||
}
|
||||
return best;
|
||||
}
|
||||
|
||||
// 目标布局里离 desired 最近、且在 0.05 容差内的输出声道;没有返回 -1。
|
||||
int nearest_channel(const double desired[3], const std::int32_t positions[][3],
|
||||
std::uint32_t channel_count) {
|
||||
int best = -1;
|
||||
double best_distance = 0.0;
|
||||
bool have_best = false;
|
||||
for (std::uint32_t index = 0; index < channel_count; ++index) {
|
||||
if (is_sentinel(positions[index])) {
|
||||
continue;
|
||||
}
|
||||
const double point[3] = {
|
||||
static_cast<double>(positions[index][0]) / 32768.0,
|
||||
static_cast<double>(positions[index][1]) / 32768.0,
|
||||
static_cast<double>(positions[index][2]) / 32768.0,
|
||||
};
|
||||
const double distance = render_distance(desired, point);
|
||||
if (!have_best || distance < best_distance) {
|
||||
best = static_cast<int>(index);
|
||||
best_distance = distance;
|
||||
have_best = true;
|
||||
}
|
||||
}
|
||||
if (have_best && best_distance <= 0.05) {
|
||||
return best;
|
||||
}
|
||||
return -1;
|
||||
}
|
||||
|
||||
// 输出声道 index -> 该声道的 Q15 方位;LFE 与非空间声道给 sentinel。
|
||||
void channel_positions(const speaker::LayoutInfo& layout,
|
||||
std::int32_t positions[][3]) {
|
||||
for (int index = 0; index < kMaxChannels; ++index) {
|
||||
set_sentinel(positions[index]);
|
||||
}
|
||||
const auto* geometry = ejoc::speaker_tables::find_layout(layout.speaker_bitfield);
|
||||
if (geometry != nullptr) {
|
||||
bool known[kMaxChannels] = {};
|
||||
std::int32_t found[kMaxChannels][3];
|
||||
for (int index = 0; index < kMaxChannels; ++index) {
|
||||
set_sentinel(found[index]);
|
||||
}
|
||||
for (const auto& region : geometry->regions) {
|
||||
for (std::size_t point = 0; point < region.point_count; ++point) {
|
||||
const auto speaker = static_cast<int>(region.points[point].speaker_id);
|
||||
if (speaker < 0 || speaker >= kMaxChannels || known[speaker]) {
|
||||
continue;
|
||||
}
|
||||
for (int axis = 0; axis < 3; ++axis) {
|
||||
found[speaker][axis] =
|
||||
static_cast<std::int32_t>(region.points[point].coordinate_q15[axis]);
|
||||
}
|
||||
known[speaker] = true;
|
||||
}
|
||||
}
|
||||
const std::uint32_t count =
|
||||
layout.channel_count < static_cast<std::uint32_t>(kMaxChannels)
|
||||
? layout.channel_count
|
||||
: static_cast<std::uint32_t>(kMaxChannels);
|
||||
for (std::uint32_t index = 0; index < count; ++index) {
|
||||
const char* name = layout.channel_names != nullptr ? layout.channel_names[index]
|
||||
: nullptr;
|
||||
if (name != nullptr && std::strcmp(name, "LFE") == 0) {
|
||||
continue;
|
||||
}
|
||||
if (known[index]) {
|
||||
for (int axis = 0; axis < 3; ++axis) {
|
||||
positions[index][axis] = found[index][axis];
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
int channel_index(const speaker::LayoutInfo& layout, const char* name) {
|
||||
if (name == nullptr || layout.channel_names == nullptr) {
|
||||
return -1;
|
||||
}
|
||||
for (std::uint32_t index = 0; index < layout.channel_count; ++index) {
|
||||
if (std::strcmp(layout.channel_names[index], name) == 0) {
|
||||
return static_cast<int>(index);
|
||||
}
|
||||
}
|
||||
return -1;
|
||||
}
|
||||
|
||||
double attenuation_db(const speaker::LayoutInfo& layout) {
|
||||
for (const AttenuationEntry& entry : kAttenuation) {
|
||||
if (std::strcmp(entry.compact_name, layout.compact_name) == 0) {
|
||||
return entry.db;
|
||||
}
|
||||
}
|
||||
return 0.0;
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
const std::int32_t (*bed_reference_positions_q15())[3] {
|
||||
return kReferencePositions;
|
||||
}
|
||||
|
||||
BedOutputDescription bed_output_description(const speaker::LayoutInfo& layout) {
|
||||
BedOutputDescription description;
|
||||
std::int32_t positions[kMaxChannels][3];
|
||||
channel_positions(layout, positions);
|
||||
const std::uint32_t output_count =
|
||||
layout.channel_count < static_cast<std::uint32_t>(kMaxChannels)
|
||||
? layout.channel_count
|
||||
: static_cast<std::uint32_t>(kMaxChannels);
|
||||
description.output_labels.reserve(output_count);
|
||||
for (std::uint32_t index = 0; index < output_count; ++index) {
|
||||
description.output_labels.push_back(nearest_reference(positions[index]));
|
||||
}
|
||||
for (int label = 0; label < oamd::kBedLabelCount; ++label) {
|
||||
const BedChannel channel = static_cast<BedChannel>(label);
|
||||
const char* name = channel == BedChannel::kLfe2 ? nullptr
|
||||
: bed_channel_short_name(channel);
|
||||
int slot = -1;
|
||||
if (name != nullptr && name[0] != '\0') {
|
||||
slot = channel_index(layout, name);
|
||||
}
|
||||
if (slot < 0 && channel == BedChannel::kLfe2) {
|
||||
slot = channel_index(layout, "LFE");
|
||||
}
|
||||
const TableB10Position* position = table_b10(channel);
|
||||
double desired[3] = {0.0, 0.0, 0.0};
|
||||
if (position != nullptr) {
|
||||
render_coordinates(position->xyz, desired);
|
||||
}
|
||||
if (slot < 0 && position != nullptr) {
|
||||
slot = nearest_channel(desired, positions, output_count);
|
||||
}
|
||||
if (slot < 0) {
|
||||
description.label_to_slot[label] = is_lfe_label(channel) ? -1 : -2;
|
||||
} else {
|
||||
description.label_to_slot[label] = slot;
|
||||
}
|
||||
if (is_lfe_label(channel)) {
|
||||
description.direct_output_labels[label] = 3;
|
||||
} else {
|
||||
int best = 3;
|
||||
double best_distance = 0.0;
|
||||
bool have_best = false;
|
||||
for (int index = 0; index < kBedReferencePositionCount; ++index) {
|
||||
const std::int32_t* candidate = kReferencePositions[index];
|
||||
if (candidate[0] == -1 && candidate[1] == -1 && candidate[2] == -1) {
|
||||
continue;
|
||||
}
|
||||
const double point[3] = {
|
||||
static_cast<double>(candidate[0]) / 32768.0,
|
||||
static_cast<double>(candidate[1]) / 32768.0,
|
||||
static_cast<double>(candidate[2]) / 32768.0,
|
||||
};
|
||||
const double distance = render_distance(desired, point);
|
||||
if (!have_best || distance < best_distance) {
|
||||
best = index;
|
||||
best_distance = distance;
|
||||
have_best = true;
|
||||
}
|
||||
}
|
||||
description.direct_output_labels[label] = best;
|
||||
}
|
||||
}
|
||||
description.automatic_attenuation_db = attenuation_db(layout);
|
||||
return description;
|
||||
}
|
||||
|
||||
} // namespace joc::joc
|
||||
@@ -6,31 +6,12 @@
|
||||
#include <vector>
|
||||
|
||||
#include "oamd/oamd_parser.h"
|
||||
#include "speaker/speaker_layout_lookup.h"
|
||||
|
||||
namespace joc::joc {
|
||||
|
||||
// OAMD Table 12 的标准 bed 声道。LSB 顺序与声明顺序一致:L/R、C、LFE、
|
||||
// Ls/Rs、Lrs/Rrs、TFL/TFR、TSL/TSR、TBL/TBR、LW/RW、LFE2。
|
||||
enum class BedChannel : std::uint8_t {
|
||||
kLeft = 0,
|
||||
kRight,
|
||||
kCenter,
|
||||
kLfe,
|
||||
kLeftSurround,
|
||||
kRightSurround,
|
||||
kLeftRearSurround,
|
||||
kRightRearSurround,
|
||||
kTopFrontLeft,
|
||||
kTopFrontRight,
|
||||
kTopSideLeft,
|
||||
kTopSideRight,
|
||||
kTopBackLeft,
|
||||
kTopBackRight,
|
||||
kLeftWide,
|
||||
kRightWide,
|
||||
kLfe2,
|
||||
kCount,
|
||||
};
|
||||
// BedChannel lives in oamd/oamd_parser.h: the enum is the OAMD Table 12
|
||||
// label order and OAMD frame state carries it directly.
|
||||
|
||||
// 内容声明的 bed 布局。bed 对象在 OAMD 里没有位置字段,位置由这里的布局决定。
|
||||
struct BedLayout {
|
||||
@@ -61,8 +42,28 @@ bool bed_channel_adm_position(BedChannel channel, double xyz[3]);
|
||||
bool bed_channel_q15(BedChannel channel, std::uint16_t q15[3]);
|
||||
|
||||
// 目标布局里同名扬声器的 Q15 方位:bed 声道直连到该扬声器,而不是参与声像。
|
||||
// 目标布局没有这个声道时返回 false。
|
||||
// 目标布局没有这个声道时返回 false。4.0 与 6.0 共用未分配的 out_ch_config
|
||||
// 0xFF,因此这里按 4.0 的声道集合解释 0xFF。
|
||||
bool layout_bed_anchor(std::uint32_t out_ch_config, BedChannel channel,
|
||||
std::uint16_t q15[3]);
|
||||
|
||||
// 参考位置网格 ``BED_REFERENCE_POSITIONS_Q15``:bed 标签与输出 slots 的最近邻
|
||||
// 匹配都用这一组固定 Q15 点。索引 3 是 (-1,-1,-1),表示 LFE 那样的非空间声源。
|
||||
inline constexpr int kBedReferencePositionCount = 38;
|
||||
const std::int32_t (*bed_reference_positions_q15())[3];
|
||||
|
||||
// 一个目标扬声器布局的 bed 路由描述(bed_layout.bed_output_description):
|
||||
// 按标准 bed 标签序(Table 12 的 17 个标签)给出目标 layout 的输出参考位置、
|
||||
// 每个标签的输出 slot 与直连参考位置,以及该布局的自动 trim 基准衰减。
|
||||
struct BedOutputDescription {
|
||||
std::vector<int> output_labels; // 每个输出声道最近邻的参考位置索引
|
||||
int label_to_slot[oamd::kBedLabelCount] = {}; // -2 不可达,-1 LFE 通路
|
||||
int direct_output_labels[oamd::kBedLabelCount] = {};
|
||||
double automatic_attenuation_db = 0.0;
|
||||
};
|
||||
|
||||
// 由布局名/几何派生 bed 路由描述;``layout.channel_names`` 必须按标准声道
|
||||
// 顺序给出 ``layout.channel_count`` 个名字。
|
||||
BedOutputDescription bed_output_description(const speaker::LayoutInfo& layout);
|
||||
|
||||
} // namespace joc::joc
|
||||
@@ -182,6 +182,77 @@ std::int16_t num_bands_for_index(std::uint32_t num_bands_idx) {
|
||||
return -1;
|
||||
}
|
||||
|
||||
namespace {
|
||||
|
||||
// Reads the 12-bit joc_header at the start of the payload. Mirrors
|
||||
// parse_joc_header(): a downmix configuration outside 0..4 is a frame status,
|
||||
// not an exception, because a metadata scan records it as an unsupported
|
||||
// variant while the frame itself stays classified.
|
||||
Status parse_id14_header(const std::uint8_t* payload, std::size_t payload_size,
|
||||
joc_frame_params* out) {
|
||||
bits::BitReader reader(payload, payload_size);
|
||||
out->dmx_config_idx = static_cast<std::uint8_t>(reader.read(3));
|
||||
out->num_objects_bits = static_cast<std::uint8_t>(reader.read(6));
|
||||
out->ext_config_idx = static_cast<std::uint8_t>(reader.read(3));
|
||||
if (reader.failed()) {
|
||||
return truncated_fail(reader);
|
||||
}
|
||||
const std::int16_t n_channels = num_channels_for_config(out->dmx_config_idx);
|
||||
if (n_channels < 0) {
|
||||
// The core channel count decides how the object syntax expands, so an
|
||||
// unknown configuration stops the parse after the header fields.
|
||||
out->status = JOC_FRAME_STATUS_UNSUPPORTED_DMX;
|
||||
out->objects_clamped = 0u;
|
||||
out->n_objects = 0u;
|
||||
return Status::success();
|
||||
}
|
||||
out->n_channels = static_cast<std::uint8_t>(n_channels);
|
||||
const std::uint32_t declared = static_cast<std::uint32_t>(out->num_objects_bits) + 1u;
|
||||
out->objects_clamped = declared > JOC_MAX_OBJECTS ? 1u : 0u;
|
||||
out->n_objects = static_cast<std::uint8_t>(
|
||||
std::min<std::uint32_t>(declared, static_cast<std::uint32_t>(JOC_MAX_OBJECTS)));
|
||||
if (out->objects_clamped != 0u) {
|
||||
out->status = JOC_FRAME_STATUS_OBJECTS_CLAMPED;
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
// Records where joc_data ends and the first trailing bytes, like _tail_info().
|
||||
void tail_info(joc_frame_params* out, const bits::BitReader& reader, const std::uint8_t* payload,
|
||||
std::size_t payload_size) {
|
||||
out->data_end_bits = static_cast<std::uint32_t>(reader.position());
|
||||
out->trailing_bits = static_cast<std::uint32_t>(payload_size * 8u - reader.position());
|
||||
const std::size_t tail_offset = reader.position() / 8u;
|
||||
if (tail_offset < payload_size) {
|
||||
const std::size_t tail_bytes = std::min<std::size_t>(8u, payload_size - tail_offset);
|
||||
std::memcpy(out->tail_bytes, payload + tail_offset, tail_bytes);
|
||||
}
|
||||
}
|
||||
|
||||
// A payload that runs out of bits after the joc_header is a damaged but
|
||||
// identifiable JOC frame: the header, clipgain, sequence counter and object
|
||||
// parameters are intact, so the frame keeps its identity and the renderer can
|
||||
// hold the previous matrices instead of stopping the stream. A payload that
|
||||
// ends before the header cannot be classified and stays a transport error; a
|
||||
// Huffman walk that leaves the tree is a syntax error, never a truncation.
|
||||
Status truncated_frame(const bits::BitReader& reader, joc_frame_params* out,
|
||||
const std::uint8_t* payload, std::size_t payload_size) {
|
||||
if (reader.error() != JOC_ERR_BITSTREAM_TRUNCATED) {
|
||||
return syntax_fail(reader.error_message());
|
||||
}
|
||||
out->status = JOC_FRAME_STATUS_TRUNCATED;
|
||||
out->present_mask = 0u;
|
||||
for (std::uint32_t obj = 0; obj < out->n_objects; ++obj) {
|
||||
if (out->objects[obj].present != 0u) {
|
||||
out->present_mask |= 1u << obj;
|
||||
}
|
||||
}
|
||||
tail_info(out, reader, payload, payload_size);
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
Status parse_id14(const std::uint8_t* payload, std::size_t payload_size, joc_frame_params* out,
|
||||
FrameSymbols* symbols) {
|
||||
if (payload == nullptr || out == nullptr) {
|
||||
@@ -195,45 +266,53 @@ Status parse_id14(const std::uint8_t* payload, std::size_t payload_size, joc_fra
|
||||
FrameSymbols& capture = (symbols != nullptr) ? *symbols : local_symbols;
|
||||
std::memset(&capture, 0, sizeof(capture));
|
||||
|
||||
bits::BitReader reader(payload, payload_size);
|
||||
|
||||
out->dmx_config_idx = static_cast<std::uint8_t>(reader.read(3));
|
||||
out->num_objects_bits = static_cast<std::uint8_t>(reader.read(6));
|
||||
out->ext_config_idx = static_cast<std::uint8_t>(reader.read(3));
|
||||
|
||||
const std::uint32_t declared_objects = static_cast<std::uint32_t>(out->num_objects_bits) + 1u;
|
||||
const std::int16_t n_channels = num_channels_for_config(out->dmx_config_idx);
|
||||
if (n_channels < 0) {
|
||||
return syntax_fail("unknown JOC downmix configuration " +
|
||||
std::to_string(out->dmx_config_idx));
|
||||
const Status header = parse_id14_header(payload, payload_size, out);
|
||||
if (!header.ok()) {
|
||||
return header;
|
||||
}
|
||||
out->n_channels = static_cast<std::uint8_t>(n_channels);
|
||||
// A declared object count above 15 is clamped to 15: the frame is parsed for
|
||||
// 15 objects only, and the clamp is reported separately.
|
||||
out->objects_clamped = declared_objects > JOC_MAX_OBJECTS ? 1u : 0u;
|
||||
const std::uint32_t n_objects =
|
||||
std::min<std::uint32_t>(declared_objects, JOC_MAX_OBJECTS);
|
||||
out->n_objects = static_cast<std::uint8_t>(n_objects);
|
||||
|
||||
bits::BitReader reader(payload, payload_size);
|
||||
reader.skip(12);
|
||||
out->clipgain_x_bits = static_cast<std::uint8_t>(reader.read(3));
|
||||
out->clipgain_y_bits = static_cast<std::uint8_t>(reader.read(5));
|
||||
out->seq_count = reader.read(10);
|
||||
// clipgain = 1 + (y/32) * 2^(x-4); ldexp builds the power of two exactly
|
||||
// over the whole legal range.
|
||||
out->clipgain = 1.0 + static_cast<double>(out->clipgain_y_bits) / 32.0 *
|
||||
std::ldexp(1.0, static_cast<int>(out->clipgain_x_bits) - 4);
|
||||
if (reader.failed()) {
|
||||
return truncated_frame(reader, out, payload, payload_size);
|
||||
}
|
||||
// clipgain = 2^(x-4) * (1 + y/32); ldexp builds the power of two exactly
|
||||
// over the whole legal range, so the result is bit-exact to the reference.
|
||||
out->clipgain = std::ldexp(1.0, static_cast<int>(out->clipgain_x_bits) - 4) *
|
||||
(1.0 + static_cast<double>(out->clipgain_y_bits) / 32.0);
|
||||
|
||||
if (out->status == JOC_FRAME_STATUS_UNSUPPORTED_DMX) {
|
||||
tail_info(out, reader, payload, payload_size);
|
||||
return Status::success();
|
||||
}
|
||||
const std::int16_t n_channels = num_channels_for_config(out->dmx_config_idx);
|
||||
const std::uint32_t n_objects = out->n_objects;
|
||||
|
||||
for (std::uint32_t obj = 0; obj < n_objects; ++obj) {
|
||||
joc_object_params& info = out->objects[obj];
|
||||
info.present = static_cast<std::uint8_t>(reader.read(1));
|
||||
if (reader.failed()) {
|
||||
return truncated_frame(reader, out, payload, payload_size);
|
||||
}
|
||||
if (info.present == 0u) {
|
||||
continue;
|
||||
}
|
||||
info.num_bands_idx = static_cast<std::uint8_t>(reader.read(3));
|
||||
if (reader.failed()) {
|
||||
return truncated_frame(reader, out, payload, payload_size);
|
||||
}
|
||||
const std::int16_t bands = num_bands_for_index(info.num_bands_idx);
|
||||
if (bands < 0) {
|
||||
return syntax_fail("unknown JOC num_bands index " +
|
||||
std::to_string(info.num_bands_idx));
|
||||
// The symbol count depends on the parameter-band count, so the rest
|
||||
// of the payload cannot be parsed.
|
||||
if (out->status == JOC_FRAME_STATUS_OK) {
|
||||
out->status = JOC_FRAME_STATUS_BAD_PARAMETER_BAND;
|
||||
}
|
||||
tail_info(out, reader, payload, payload_size);
|
||||
return Status::success();
|
||||
}
|
||||
info.n_bands = static_cast<std::uint8_t>(bands);
|
||||
info.sparse = static_cast<std::uint8_t>(reader.read(1));
|
||||
@@ -241,16 +320,42 @@ Status parse_id14(const std::uint8_t* payload, std::size_t payload_size, joc_fra
|
||||
info.slope_idx = static_cast<std::uint8_t>(reader.read(1));
|
||||
info.num_dpoints_bits = static_cast<std::uint8_t>(reader.read(1));
|
||||
info.n_dpoints = static_cast<std::uint8_t>(info.num_dpoints_bits + 1u);
|
||||
if (reader.failed()) {
|
||||
return truncated_frame(reader, out, payload, payload_size);
|
||||
}
|
||||
if (info.slope_idx == 1u) {
|
||||
for (std::uint32_t dp = 0; dp < info.n_dpoints; ++dp) {
|
||||
info.offset_ts[dp] = static_cast<std::uint8_t>(reader.read(5) + 1u);
|
||||
}
|
||||
}
|
||||
info.offset_ts[dp] = static_cast<std::uint8_t>(reader.read(5));
|
||||
}
|
||||
if (reader.failed()) {
|
||||
return truncated_fail(reader);
|
||||
return truncated_frame(reader, out, payload, payload_size);
|
||||
}
|
||||
// offset_ts is the raw 5-bit slot index (legal range 0..23) and one
|
||||
// object's data points must be strictly increasing.
|
||||
bool invalid = false;
|
||||
for (std::uint32_t dp = 0; dp < info.n_dpoints; ++dp) {
|
||||
if (info.offset_ts[dp] >= JOC_TIMESLOTS) {
|
||||
invalid = true;
|
||||
break;
|
||||
}
|
||||
}
|
||||
if (!invalid && info.n_dpoints > 1u) {
|
||||
for (std::uint32_t dp = 1; dp < info.n_dpoints; ++dp) {
|
||||
if (info.offset_ts[dp] <= info.offset_ts[dp - 1u]) {
|
||||
invalid = true;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
if (invalid && out->status == JOC_FRAME_STATUS_OK) {
|
||||
out->status = JOC_FRAME_STATUS_BAD_OFFSET_TS;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// joc_data (Huffman): dense reads MTX per channel and band, sparse reads IDX
|
||||
// then VEC. A frame whose status is already non-zero dequantizes below but
|
||||
// is never rendered, so a misaligned parse cannot reach the DSP.
|
||||
for (std::uint32_t obj = 0; obj < n_objects; ++obj) {
|
||||
const joc_object_params& info = out->objects[obj];
|
||||
if (info.present == 0u) {
|
||||
@@ -267,13 +372,13 @@ Status parse_id14(const std::uint8_t* payload, std::size_t payload_size, joc_fra
|
||||
const Tree idx_tree = select_tree(info.quant_idx, SymbolKind::Idx, n_channels);
|
||||
const std::uint32_t first = reader.read(3);
|
||||
if (reader.failed()) {
|
||||
return truncated_fail(reader);
|
||||
return truncated_frame(reader, out, payload, payload_size);
|
||||
}
|
||||
symbol->idx[dp][0] = static_cast<std::uint8_t>(first);
|
||||
for (int pb = 1; pb < info.n_bands; ++pb) {
|
||||
std::int16_t value = 0;
|
||||
if (!huff_decode(idx_tree, reader, &value)) {
|
||||
return truncated_fail(reader);
|
||||
return truncated_frame(reader, out, payload, payload_size);
|
||||
}
|
||||
symbol->idx[dp][pb] = static_cast<std::uint8_t>(value);
|
||||
}
|
||||
@@ -281,7 +386,7 @@ Status parse_id14(const std::uint8_t* payload, std::size_t payload_size, joc_fra
|
||||
for (int pb = 0; pb < info.n_bands; ++pb) {
|
||||
std::int16_t value = 0;
|
||||
if (!huff_decode(vec_tree, reader, &value)) {
|
||||
return truncated_fail(reader);
|
||||
return truncated_frame(reader, out, payload, payload_size);
|
||||
}
|
||||
symbol->vec[dp][pb] = value;
|
||||
}
|
||||
@@ -291,7 +396,7 @@ Status parse_id14(const std::uint8_t* payload, std::size_t payload_size, joc_fra
|
||||
for (int pb = 0; pb < info.n_bands; ++pb) {
|
||||
std::int16_t value = 0;
|
||||
if (!huff_decode(mtx_tree, reader, &value)) {
|
||||
return truncated_fail(reader);
|
||||
return truncated_frame(reader, out, payload, payload_size);
|
||||
}
|
||||
symbol->mtx[dp][ch][pb] = value;
|
||||
}
|
||||
@@ -300,12 +405,16 @@ Status parse_id14(const std::uint8_t* payload, std::size_t payload_size, joc_fra
|
||||
}
|
||||
}
|
||||
|
||||
out->data_end_bits = static_cast<std::uint32_t>(reader.position());
|
||||
out->trailing_bits = static_cast<std::uint32_t>(payload_size * 8u - reader.position());
|
||||
const std::size_t tail_offset = reader.position() / 8u;
|
||||
if (tail_offset < payload_size) {
|
||||
const std::size_t tail_bytes = std::min<std::size_t>(8u, payload_size - tail_offset);
|
||||
std::memcpy(out->tail_bytes, payload + tail_offset, tail_bytes);
|
||||
tail_info(out, reader, payload, payload_size);
|
||||
if (out->status == JOC_FRAME_STATUS_OK && out->ext_config_idx == 0u) {
|
||||
// The tail pads with zero bits up to the next byte; a whole byte after
|
||||
// that alignment is unexpected data.
|
||||
const std::uint32_t pad =
|
||||
(8u - (static_cast<std::uint32_t>(reader.position()) % 8u)) % 8u;
|
||||
const std::uint32_t total_bits = static_cast<std::uint32_t>(payload_size * 8u);
|
||||
if (total_bits > static_cast<std::uint32_t>(reader.position()) + pad) {
|
||||
out->status = JOC_FRAME_STATUS_TRAILING_DATA;
|
||||
}
|
||||
}
|
||||
|
||||
for (std::uint32_t obj = 0; obj < n_objects; ++obj) {
|
||||
@@ -313,8 +422,7 @@ Status parse_id14(const std::uint8_t* payload, std::size_t payload_size, joc_fra
|
||||
if (info.present == 0u) {
|
||||
continue;
|
||||
}
|
||||
std::uint32_t mask_bit = 1u << obj;
|
||||
out->present_mask |= mask_bit;
|
||||
out->present_mask |= 1u << obj;
|
||||
|
||||
const std::int64_t nquant = (info.quant_idx == 0u) ? 96 : 192;
|
||||
std::int64_t q[JOC_MAX_DPOINTS][JOC_MAX_CORE_CHANNELS][JOC_MAX_PARAMETER_BANDS] = {};
|
||||
@@ -372,7 +480,16 @@ int score_container_as_joc(const std::uint8_t* data, std::size_t size,
|
||||
if (emdf::extract_payload_bytes(data, size, *payload, &bytes).ok()) {
|
||||
joc_frame_params params;
|
||||
FrameSymbols symbols;
|
||||
parses = parse_id14(bytes.data(), bytes.size(), ¶ms, &symbols).ok();
|
||||
if (parse_id14(bytes.data(), bytes.size(), ¶ms, &symbols).ok()) {
|
||||
// The reference scores a payload by the frame status: an unsupported
|
||||
// downmix configuration (2), an unknown parameter-band count (4) or a
|
||||
// payload that ends inside joc_data (7) must not win the container
|
||||
// choice, because another candidate may carry the frame's complete
|
||||
// metadata.
|
||||
parses = params.status != JOC_FRAME_STATUS_UNSUPPORTED_DMX &&
|
||||
params.status != JOC_FRAME_STATUS_BAD_PARAMETER_BAND &&
|
||||
params.status != JOC_FRAME_STATUS_TRUNCATED;
|
||||
}
|
||||
}
|
||||
const bool associated = oamd->has_group_id != 0u && payload->has_group_id != 0u &&
|
||||
oamd->group_id == payload->group_id;
|
||||
@@ -406,7 +523,20 @@ Status parse_eac3_frame(const std::uint8_t* frame, std::size_t frame_size, joc_f
|
||||
if (!extract.ok()) {
|
||||
return extract;
|
||||
}
|
||||
return parse_id14(bytes.data(), bytes.size(), out, symbols);
|
||||
const Status parsed = parse_id14(bytes.data(), bytes.size(), out, symbols);
|
||||
if (!parsed.ok()) {
|
||||
return parsed;
|
||||
}
|
||||
// A metadata scan rejects an unsupported downmix configuration outright: the
|
||||
// core channel count decides how every later frame is expanded, so this is a
|
||||
// variant error rather than a per-frame status.
|
||||
if (out->status == JOC_FRAME_STATUS_UNSUPPORTED_DMX) {
|
||||
const std::string message =
|
||||
"downmix_config_unsupported: ID14 declares joc_dmx_config_idx=" +
|
||||
std::to_string(out->dmx_config_idx) + " (legal values are 0..4)";
|
||||
return Status::fail(JOC_ERR_JOC_UNSUPPORTED_VARIANT, stage::kJoc, message);
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
Status check_id14_padding(const std::uint8_t* payload, std::size_t payload_size,
|
||||
@@ -417,6 +547,21 @@ Status check_id14_padding(const std::uint8_t* payload, std::size_t payload_size,
|
||||
if (!status.ok()) {
|
||||
return status;
|
||||
}
|
||||
// The statuses that stop the parse early leave the payload without a
|
||||
// joc_data end position, so they keep reporting the failure they reported
|
||||
// before the frame-status split.
|
||||
if (params.status == JOC_FRAME_STATUS_UNSUPPORTED_DMX) {
|
||||
return Status::fail(JOC_ERR_JOC_UNSUPPORTED_VARIANT, stage::kJoc,
|
||||
"downmix_config_unsupported: joc_dmx_config_idx outside 0..4");
|
||||
}
|
||||
if (params.status == JOC_FRAME_STATUS_BAD_PARAMETER_BAND) {
|
||||
return Status::fail(JOC_ERR_JOC_SYNTAX, stage::kJoc,
|
||||
"unknown JOC num_bands index stops the payload");
|
||||
}
|
||||
if (params.status == JOC_FRAME_STATUS_TRUNCATED) {
|
||||
return Status::fail(JOC_ERR_BITSTREAM_TRUNCATED, stage::kJoc,
|
||||
"the ID14 payload ends before its joc_data is complete");
|
||||
}
|
||||
if (out_trailing_bits != nullptr) {
|
||||
*out_trailing_bits = params.trailing_bits;
|
||||
}
|
||||
@@ -861,6 +1006,7 @@ Status check_cross_layer(const joc_frame_params& params, const eac3::Frame* core
|
||||
std::uint16_t chanmap,
|
||||
const std::vector<oamd::BedAssignment>* bed_assignments,
|
||||
std::string* error, CrossLayerNotes* notes) {
|
||||
(void)has_dependent;
|
||||
const std::uint8_t config_idx = params.dmx_config_idx;
|
||||
const int core_channels = params.n_channels;
|
||||
std::vector<Position> wanted;
|
||||
|
||||
+563
-30
@@ -69,6 +69,19 @@ public:
|
||||
std::memset(synthesis_state_, 0, sizeof(synthesis_state_));
|
||||
std::memset(x_, 0, sizeof(x_));
|
||||
std::memset(z_, 0, sizeof(z_));
|
||||
// Frame-level state: per-object previous matrices and presence markers,
|
||||
// the seq state machine, the 8-plane passthrough history, the LFE ramp
|
||||
// start value and the previous actual core-channel count.
|
||||
std::memset(prev_matrix_, 0, sizeof(prev_matrix_));
|
||||
std::memset(prev_present_, 0, sizeof(prev_present_));
|
||||
std::memset(core_history_, 0, sizeof(core_history_));
|
||||
state_ = 0;
|
||||
expected_seq_ = 0;
|
||||
invalid_count_ = 0;
|
||||
last_frame_valid_ = false;
|
||||
n_channels_prev_ = -1;
|
||||
has_lfe_clipgain_ = false;
|
||||
lfe_clipgain_ = 0.0;
|
||||
// Sentinel: the first frame after reset takes the new scalar directly,
|
||||
// without an intra-frame transition.
|
||||
analysis_phase_ = -0.5f;
|
||||
@@ -174,13 +187,22 @@ public:
|
||||
}
|
||||
|
||||
std::memset(output16, 0, sizeof(float) * EJOC_OUTPUT_CHANNELS * EJOC_FRAME_SAMPLES);
|
||||
analysis(bed_planar, phase_new, clipgain);
|
||||
render_lfe(lfe, output_scale, output16);
|
||||
// Analysis-side scalar = the structural 1/16 times this frame's clipgain.
|
||||
// clipgain enters the analysis at this stage and the x16 at the output
|
||||
// cancels the 1/16, so the net effect is a multiply by joc_clipgain.
|
||||
analysis(bed_planar, static_cast<float>(phase_new * clipgain));
|
||||
decode_lfe(lfe, clipgain, output16, output_scale);
|
||||
process_objects(object_mask, n_bands, n_dpoints, slope_idx, offset_ts,
|
||||
dq, output_scale, output16);
|
||||
return 0;
|
||||
}
|
||||
|
||||
// Frame-level entry point; the state machine and per-object history live
|
||||
// inside the renderer (see render_frame_impl()).
|
||||
int render_frame(const ejoc_frame_input& input) noexcept {
|
||||
return render_frame_impl(input);
|
||||
}
|
||||
|
||||
private:
|
||||
void initialize_tables() noexcept {
|
||||
for (int i = 0; i < 64; ++i) {
|
||||
@@ -378,13 +400,12 @@ private:
|
||||
}
|
||||
}
|
||||
|
||||
void analysis(const float* bed_planar, const float phase_new, const double clipgain) noexcept {
|
||||
// Frames the analysis QMF of the core PCM with the given analysis-side
|
||||
// scalar (the structural 1/16 times joc_clipgain). The scalar ramps over the
|
||||
// frame's first 256 samples from the previous frame's value; the first frame
|
||||
// after a reset takes it directly.
|
||||
void analysis(const float* bed_planar, const float scalar) noexcept {
|
||||
job_bed_ = bed_planar;
|
||||
// Analysis-side scalar = the structural 1/16 times this frame's clipgain.
|
||||
// clipgain enters the analysis at this stage and the x16 at the output
|
||||
// cancels the 1/16, so the net effect is a multiply by joc_clipgain; the
|
||||
// previous frame's scalar is the ramp start.
|
||||
const float scalar = static_cast<float>(phase_new * clipgain);
|
||||
job_phase_old_ = analysis_phase_;
|
||||
if (job_phase_old_ <= 0.0f && job_phase_old_ != 0.0f) {
|
||||
job_phase_old_ = scalar;
|
||||
@@ -509,11 +530,11 @@ private:
|
||||
}
|
||||
qmf5_step(state, rotated, pcm64);
|
||||
for (int sample = 0; sample < 64; ++sample) {
|
||||
// The clamp is on the ×16 object PCM; joc_clipgain already
|
||||
// entered the analysis, so only the user output gain remains, and
|
||||
// it is a float32 multiply like the reference.
|
||||
const double clipped = clamp_unit(16.0 * pcm64[sample]);
|
||||
// joc_clipgain already entered the analysis (see analysis()); what
|
||||
// remains here is the user output gain.
|
||||
const float value = static_cast<float>(clipped * output_scale);
|
||||
destination[timeslot * 64 + sample] = value;
|
||||
destination[timeslot * 64 + sample] = static_cast<float>(clipped) * output_scale;
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -533,8 +554,10 @@ private:
|
||||
const int item = job_objects_[index];
|
||||
if (job_kind_ == JobKind::Analysis) {
|
||||
analysis_channel(item);
|
||||
} else {
|
||||
} else if (job_kind_ == JobKind::Objects) {
|
||||
process_one_object(item);
|
||||
} else {
|
||||
process_one_object_frame(item);
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -620,24 +643,503 @@ private:
|
||||
dispatch_jobs();
|
||||
}
|
||||
|
||||
void render_lfe(const float* lfe, const float output_scale, float* output16) noexcept {
|
||||
if (!lfe) {
|
||||
return;
|
||||
// Core LFE: the full joc_clipgain is applied as a sample-wise ramp before the
|
||||
// 1217-sample delay, and the read-out is only clamped. The ramp is
|
||||
// G_old + n*(G_new-G_old)/256 for n < 256 and G_new afterwards; the first
|
||||
// frame after a reset uses G_old = G_new.
|
||||
void decode_lfe(const float* lfe, const double clipgain, float* output16,
|
||||
const float output_scale) noexcept {
|
||||
const double new_gain = clipgain;
|
||||
const double old_gain = has_lfe_clipgain_ ? lfe_clipgain_ : new_gain;
|
||||
const bool ramped = old_gain != new_gain;
|
||||
const double step = (new_gain - old_gain) / 256.0;
|
||||
double ramp[256];
|
||||
for (int n = 0; n < 256; ++n) {
|
||||
ramp[n] = ramped ? old_gain + static_cast<double>(n) * step : new_gain;
|
||||
}
|
||||
has_lfe_clipgain_ = true;
|
||||
lfe_clipgain_ = new_gain;
|
||||
|
||||
float* channel0 = output16;
|
||||
for (int sample = 0; sample < kLfeDelay; ++sample) {
|
||||
channel0[sample] =
|
||||
static_cast<float>(clamp_unit(lfe_delay_[sample])) * output_scale;
|
||||
}
|
||||
for (int sample = kLfeDelay; sample < EJOC_FRAME_SAMPLES; ++sample) {
|
||||
const int input = sample - kLfeDelay;
|
||||
const double gain = input < 256 ? ramp[input] : new_gain;
|
||||
const double value = (lfe != nullptr) ? static_cast<double>(lfe[input]) : 0.0;
|
||||
channel0[sample] =
|
||||
static_cast<float>(clamp_unit(value * gain)) * output_scale;
|
||||
}
|
||||
for (int sample = 0; sample < kLfeDelay; ++sample) {
|
||||
const float value = static_cast<float>(clamp_unit(lfe_delay_[sample]));
|
||||
output16[sample] = static_cast<float>(value * output_scale);
|
||||
}
|
||||
for (int sample = kLfeDelay; sample < 1536; ++sample) {
|
||||
const float value = static_cast<float>(clamp_unit(static_cast<double>(lfe[sample - kLfeDelay])));
|
||||
output16[sample] = static_cast<float>(value * output_scale);
|
||||
}
|
||||
for (int sample = 0; sample < kLfeDelay; ++sample) {
|
||||
lfe_delay_[sample] = static_cast<double>(lfe[sample + (1536 - kLfeDelay)]);
|
||||
const int input = sample + (EJOC_FRAME_SAMPLES - kLfeDelay);
|
||||
const double gain = input < 256 ? ramp[input] : new_gain;
|
||||
const double value = (lfe != nullptr) ? static_cast<double>(lfe[input]) : 0.0;
|
||||
lfe_delay_[sample] = value * gain;
|
||||
}
|
||||
}
|
||||
|
||||
enum class JobKind : uint8_t { Analysis, Objects };
|
||||
enum class FrameMode : uint8_t { Normal, Hold, HardCut, Passthrough };
|
||||
|
||||
// ---- frame-level state machine ------------------------------------- //
|
||||
// The Renderer owns the seq state machine (normal / hold / hard cut /
|
||||
// passthrough), the per-object previous matrix and its presence marker, the
|
||||
// analysis/synthesis state, the LFE delay ring with its clipgain ramp, and
|
||||
// the 8-plane core history of the passthrough path, so one audio frame in
|
||||
// produces the complete 16-channel object frame.
|
||||
|
||||
float held_analysis_scalar() const noexcept {
|
||||
// Sentinel: a renderer that has not seen a JOC frame yet reports the
|
||||
// structural 1/16, i.e. clipgain 1.0.
|
||||
return (analysis_phase_ <= 0.0f && analysis_phase_ != 0.0f) ? 0.0625f
|
||||
: analysis_phase_;
|
||||
}
|
||||
|
||||
double held_clipgain() const noexcept {
|
||||
return static_cast<double>(held_analysis_scalar()) * 16.0;
|
||||
}
|
||||
|
||||
// The 10-bit JOC sequence counter wraps from 1023 back to 1.
|
||||
static uint32_t next_seq(const uint32_t value) noexcept {
|
||||
return value == 1023u ? 1u : value + 1u;
|
||||
}
|
||||
|
||||
// Drops the state of the core channels a smaller layout does not use.
|
||||
void clear_unused_channels(const uint32_t n_channels) noexcept {
|
||||
if (n_channels >= EJOC_MAX_CORE_CHANNELS) {
|
||||
return;
|
||||
}
|
||||
for (uint32_t channel = n_channels; channel < EJOC_MAX_CORE_CHANNELS; ++channel) {
|
||||
std::memset(analysis_fifo_[channel], 0, sizeof(analysis_fifo_[channel]));
|
||||
}
|
||||
for (int object = 0; object < EJOC_MAX_OBJECTS; ++object) {
|
||||
for (uint32_t channel = n_channels; channel < EJOC_MAX_CORE_CHANNELS; ++channel) {
|
||||
std::memset(prev_matrix_[object][channel], 0,
|
||||
sizeof(prev_matrix_[object][channel]));
|
||||
}
|
||||
}
|
||||
if (n_channels == 5u) {
|
||||
// A 5-channel frame does not drive the extra-pair delay lanes.
|
||||
std::memset(analysis_delay_[3], 0, sizeof(analysis_delay_[3]));
|
||||
std::memset(analysis_delay_[4], 0, sizeof(analysis_delay_[4]));
|
||||
}
|
||||
}
|
||||
|
||||
// A declared core-channel-count change (renderer status 9) clears the LFE
|
||||
// ring; the clipgain ramp state is kept, so the gain does not retune.
|
||||
void maybe_clear_lfe_ring(const uint32_t status) noexcept {
|
||||
if (status == 9u) {
|
||||
std::memset(lfe_delay_, 0, sizeof(lfe_delay_));
|
||||
}
|
||||
}
|
||||
|
||||
// Advances the seq state machine for this frame and reports the state the
|
||||
// renderer was in before the frame. status 8 (no metadata) neither advances
|
||||
// the expected counter nor compares the sequence.
|
||||
uint32_t advance_state(const uint32_t status, const uint32_t seq) noexcept {
|
||||
const uint32_t pre_state = state_;
|
||||
const bool valid = status == 0u || status == 9u;
|
||||
if (status == EJOC_FRAME_STATUS_NO_METADATA && pre_state != 0u) {
|
||||
state_ = 3;
|
||||
return pre_state;
|
||||
}
|
||||
expected_seq_ = next_seq(expected_seq_);
|
||||
if (valid && seq == 0u && pre_state != 0u) {
|
||||
// The programme restarted: hard cut and drop the object history.
|
||||
state_ = 4;
|
||||
return pre_state;
|
||||
}
|
||||
uint32_t state = state_;
|
||||
if (state == 0u) {
|
||||
if (valid) {
|
||||
state = 1u;
|
||||
expected_seq_ = seq;
|
||||
}
|
||||
} else if (state == 1u) {
|
||||
if (valid) {
|
||||
if (seq != expected_seq_) {
|
||||
state = 4u;
|
||||
}
|
||||
} else {
|
||||
state = 2u;
|
||||
invalid_count_ = 1u;
|
||||
}
|
||||
} else if (state == 2u) {
|
||||
if (valid) {
|
||||
state = 1u;
|
||||
expected_seq_ = seq;
|
||||
} else if (invalid_count_ == 5u) {
|
||||
state = 3u;
|
||||
} else {
|
||||
++invalid_count_;
|
||||
}
|
||||
} else if (state == 3u) {
|
||||
if (valid) {
|
||||
state = 1u;
|
||||
expected_seq_ = seq;
|
||||
} else {
|
||||
++invalid_count_;
|
||||
}
|
||||
} else {
|
||||
// state 4: the hard cut lasted one frame; a valid frame resyncs and
|
||||
// an invalid one starts the invalid-frame count.
|
||||
if (valid) {
|
||||
state = 1u;
|
||||
expected_seq_ = seq;
|
||||
} else {
|
||||
state = 2u;
|
||||
invalid_count_ = 1u;
|
||||
}
|
||||
}
|
||||
state_ = state;
|
||||
if (status == 9u) {
|
||||
state_ = 4u;
|
||||
}
|
||||
last_frame_valid_ = valid;
|
||||
return pre_state;
|
||||
}
|
||||
|
||||
// Compatibility note for frames without JOC metadata (status 8).
|
||||
//
|
||||
// The state machine used to map status 8 to Passthrough unconditionally,
|
||||
// including in the middle of a stream. That is only meaningful before the
|
||||
// first valid JOC frame: a plain E-AC-3 stream (or a stream whose first JOC
|
||||
// frame comes late) has no object state yet, so the core planes are all
|
||||
// there is and the delayed core PCM is the right output. In the middle of
|
||||
// a JOC stream a frame without metadata is a damaged frame, not "no
|
||||
// objects": the Passthrough output replaces the object render with the core
|
||||
// planes, and the object synthesis restarts from silence on the next frame.
|
||||
// Every such frame therefore produced a step at the frame boundary and a
|
||||
// second one when the metadata came back; encoders that are not careful
|
||||
// with the JOC EMDF container emit exactly those frames.
|
||||
//
|
||||
// After the change status 8 follows the state machine like the other
|
||||
// invalid statuses: state 0 (no valid frame rendered yet) stays Passthrough,
|
||||
// and state 3 (mid-stream) selects Hold, which reuses the previous object
|
||||
// matrices and keeps the bed and object rendering continuous.
|
||||
//
|
||||
// This is a compatibility choice, not a claim about the normative reading
|
||||
// of the specification: a frame without metadata can be argued to be the
|
||||
// Passthrough case. It exists so that non-standard streams stay click-free;
|
||||
// if the reference behaviour is preferred, re-add the unconditional
|
||||
// Passthrough branch here.
|
||||
FrameMode select_mode(const uint32_t status) const noexcept {
|
||||
if (status == EJOC_FRAME_STATUS_NO_METADATA) {
|
||||
// See the note above: only a stream that has not rendered a valid
|
||||
// frame yet has nothing to hold.
|
||||
return state_ == 0u ? FrameMode::Passthrough : FrameMode::Hold;
|
||||
}
|
||||
if (state_ == 4u) {
|
||||
return FrameMode::HardCut;
|
||||
}
|
||||
if (state_ == 2u || state_ == 3u) {
|
||||
return FrameMode::Hold;
|
||||
}
|
||||
if (state_ == 1u && status == 0u) {
|
||||
return FrameMode::Normal;
|
||||
}
|
||||
return FrameMode::Passthrough;
|
||||
}
|
||||
|
||||
// JOC input order 5/7 channels (without LFE) -> 8-plane core order
|
||||
// L,R,C,LFE,Ls,Rs,(Lrs,Rrs). Used by the passthrough path and its history.
|
||||
void build_core8(const float* bed_planar, const float* lfe,
|
||||
const uint32_t n_channels) noexcept {
|
||||
std::memset(core8_, 0, sizeof(core8_));
|
||||
for (uint32_t channel = 0; channel < 3u && channel < n_channels; ++channel) {
|
||||
for (int sample = 0; sample < EJOC_FRAME_SAMPLES; ++sample) {
|
||||
core8_[channel][sample] =
|
||||
static_cast<double>(bed_planar[channel * EJOC_FRAME_SAMPLES + sample]);
|
||||
}
|
||||
}
|
||||
if (lfe != nullptr) {
|
||||
for (int sample = 0; sample < EJOC_FRAME_SAMPLES; ++sample) {
|
||||
core8_[EJOC_LFE_SLOT][sample] = static_cast<double>(lfe[sample]);
|
||||
}
|
||||
}
|
||||
const uint32_t rest = n_channels > 3u ? n_channels - 3u : 0u;
|
||||
for (uint32_t index = 0; index < rest; ++index) {
|
||||
const uint32_t channel = 3u + index;
|
||||
for (int sample = 0; sample < EJOC_FRAME_SAMPLES; ++sample) {
|
||||
core8_[4u + index][sample] =
|
||||
static_cast<double>(bed_planar[channel * EJOC_FRAME_SAMPLES + sample]);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Passthrough output: slots 0..7 take the previous frame's 8 core planes
|
||||
// delayed by 1217 samples plus this frame's leading 319 samples; slots 8..15
|
||||
// stay zero.
|
||||
void write_passthrough(float* output16, const float output_scale) const noexcept {
|
||||
for (int channel = 0; channel < EJOC_CORE_OUT_CHANNELS; ++channel) {
|
||||
float* destination = output16 + channel * EJOC_FRAME_SAMPLES;
|
||||
for (int sample = 0; sample < kLfeDelay; ++sample) {
|
||||
destination[sample] = static_cast<float>(
|
||||
core_history_[channel][EJOC_FRAME_SAMPLES - kLfeDelay + sample]) *
|
||||
output_scale;
|
||||
}
|
||||
for (int sample = kLfeDelay; sample < EJOC_FRAME_SAMPLES; ++sample) {
|
||||
destination[sample] =
|
||||
static_cast<float>(core8_[channel][sample - kLfeDelay]) * output_scale;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// Normal rendering: interpolate the per-object matrix from its previous
|
||||
// frame-end value (or hard cut when there is none) and publish objects that
|
||||
// disappeared as an all-zero matrix until they come back.
|
||||
void matrix_object_frame(const int object) noexcept {
|
||||
std::memset(z_[object], 0, sizeof(z_[object]));
|
||||
const int n_channels = static_cast<int>(core_channels_);
|
||||
const bool present = (job_object_mask_ & (1u << object)) != 0u;
|
||||
if (!present) {
|
||||
if (prev_present_[object]) {
|
||||
for (int channel = 0; channel < n_channels; ++channel) {
|
||||
std::memset(prev_matrix_[object][channel], 0,
|
||||
sizeof(prev_matrix_[object][channel]));
|
||||
}
|
||||
prev_present_[object] = true;
|
||||
}
|
||||
return;
|
||||
}
|
||||
const int bands = job_n_bands_[object];
|
||||
const int points = job_n_dpoints_[object];
|
||||
const int slope = job_slope_idx_[object];
|
||||
const uint8_t* pb_map = parameter_band_map(bands);
|
||||
const bool has_previous = prev_present_[object];
|
||||
for (int channel = 0; channel < n_channels; ++channel) {
|
||||
for (int subband = 0; subband < EJOC_SUBBANDS; ++subband) {
|
||||
const int parameter_band = pb_map[subband];
|
||||
const double target0 = job_dq_[dq_index(object, 0, channel, parameter_band)];
|
||||
const double target1 = points == 2
|
||||
? job_dq_[dq_index(object, 1, channel, parameter_band)]
|
||||
: target0;
|
||||
const double previous =
|
||||
has_previous ? prev_matrix_[object][channel][subband] : 0.0;
|
||||
double last = target0;
|
||||
for (int timeslot = 0; timeslot < EJOC_TIMESLOTS; ++timeslot) {
|
||||
double coefficient = 0.0;
|
||||
if (slope == 0) {
|
||||
if (!has_previous) {
|
||||
if (points == 1 || timeslot < EJOC_TIMESLOTS / 2) {
|
||||
// Hard cut: no fade-in from zero. With two data
|
||||
// points the second half still interpolates.
|
||||
coefficient = target0;
|
||||
} else {
|
||||
const double alpha =
|
||||
static_cast<double>(timeslot - (EJOC_TIMESLOTS / 2 - 1)) /
|
||||
static_cast<double>(EJOC_TIMESLOTS / 2);
|
||||
coefficient = target0 * (1.0 - alpha) + target1 * alpha;
|
||||
}
|
||||
} else if (points == 1) {
|
||||
const double alpha =
|
||||
static_cast<double>(timeslot + 1) /
|
||||
static_cast<double>(EJOC_TIMESLOTS);
|
||||
coefficient = previous * (1.0 - alpha) + target0 * alpha;
|
||||
} else if (timeslot < EJOC_TIMESLOTS / 2) {
|
||||
const double alpha = static_cast<double>(timeslot + 1) /
|
||||
(EJOC_TIMESLOTS / 2);
|
||||
coefficient = previous * (1.0 - alpha) + target0 * alpha;
|
||||
} else {
|
||||
const double alpha =
|
||||
static_cast<double>(timeslot - (EJOC_TIMESLOTS / 2 - 1)) /
|
||||
static_cast<double>(EJOC_TIMESLOTS / 2);
|
||||
coefficient = target0 * (1.0 - alpha) + target1 * alpha;
|
||||
}
|
||||
} else {
|
||||
// Step mode switches targets at the raw offset slots.
|
||||
const double base = has_previous ? previous : target0;
|
||||
coefficient = timeslot < job_offset_ts_[object * EJOC_MAX_DPOINTS]
|
||||
? base
|
||||
: target0;
|
||||
if (points == 2 &&
|
||||
timeslot >= job_offset_ts_[object * EJOC_MAX_DPOINTS + 1]) {
|
||||
coefficient = target1;
|
||||
}
|
||||
}
|
||||
z_[object][subband][timeslot].re +=
|
||||
x_[channel][subband][timeslot].re * coefficient;
|
||||
z_[object][subband][timeslot].im +=
|
||||
x_[channel][subband][timeslot].im * coefficient;
|
||||
last = coefficient;
|
||||
}
|
||||
prev_matrix_[object][channel][subband] = last;
|
||||
}
|
||||
}
|
||||
prev_present_[object] = true;
|
||||
}
|
||||
|
||||
// Hold rendering: all 24 slots reuse the previous frame-end matrix and the
|
||||
// previous state is not updated.
|
||||
void matrix_object_hold(const int object) noexcept {
|
||||
std::memset(z_[object], 0, sizeof(z_[object]));
|
||||
const int n_channels = static_cast<int>(core_channels_);
|
||||
for (int channel = 0; channel < n_channels; ++channel) {
|
||||
for (int subband = 0; subband < EJOC_SUBBANDS; ++subband) {
|
||||
const double previous = prev_matrix_[object][channel][subband];
|
||||
for (int timeslot = 0; timeslot < EJOC_TIMESLOTS; ++timeslot) {
|
||||
z_[object][subband][timeslot].re +=
|
||||
x_[channel][subband][timeslot].re * previous;
|
||||
z_[object][subband][timeslot].im +=
|
||||
x_[channel][subband][timeslot].im * previous;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
void process_one_object_frame(const int object) noexcept {
|
||||
if (frame_mode_ == FrameMode::Normal) {
|
||||
matrix_object_frame(object);
|
||||
} else if (frame_mode_ == FrameMode::Hold) {
|
||||
matrix_object_hold(object);
|
||||
} else {
|
||||
// Hard cut and passthrough publish silence, but every object slot is
|
||||
// still synthesised so the synthesis state stays aligned.
|
||||
std::memset(z_[object], 0, sizeof(z_[object]));
|
||||
}
|
||||
synthesis_object(object, job_output_scale_, job_output16_);
|
||||
}
|
||||
|
||||
// One E-AC-3 audio frame: validates the description, advances the state
|
||||
// machine and dispatches one of the four rendering modes.
|
||||
int render_frame_impl(const ejoc_frame_input& input) noexcept {
|
||||
error_[0] = '\0';
|
||||
if (input.struct_size != sizeof(ejoc_frame_input) ||
|
||||
input.struct_version != EJOC_FRAME_INPUT_VERSION) {
|
||||
return fail("unsupported ejoc_frame_input struct version");
|
||||
}
|
||||
if (input.bed_planar == nullptr || input.output16_planar == nullptr) {
|
||||
return fail("null bed or output pointer passed to ejoc_renderer_render_frame");
|
||||
}
|
||||
const uint32_t n_channels = input.core_channel_count;
|
||||
if (n_channels != EJOC_CORE_CHANNELS && n_channels != EJOC_MAX_CORE_CHANNELS) {
|
||||
return fail("core_channel_count must be 5 or 7");
|
||||
}
|
||||
if (!std::isfinite(input.clipgain) || !std::isfinite(input.output_scale)) {
|
||||
return fail("clipgain and output_scale must be finite");
|
||||
}
|
||||
const bool has_metadata = input.has_metadata != 0u;
|
||||
uint32_t status = input.frame_status;
|
||||
if (!has_metadata) {
|
||||
if (status != EJOC_FRAME_STATUS_NO_METADATA) {
|
||||
return fail("a frame without JOC metadata must use frame_status 8");
|
||||
}
|
||||
} else if (status != 0u && (status < 2u || status > 7u)) {
|
||||
return fail("frame_status is not a JOC parse status");
|
||||
}
|
||||
if ((input.object_mask & ~((1u << EJOC_MAX_OBJECTS) - 1u)) != 0u) {
|
||||
return fail("object_mask contains an object index above 14");
|
||||
}
|
||||
if (has_metadata && status == 0u) {
|
||||
for (int object = 0; object < EJOC_MAX_OBJECTS; ++object) {
|
||||
if ((input.object_mask & (1u << object)) == 0u) {
|
||||
continue;
|
||||
}
|
||||
if (parameter_band_map(input.n_bands[object]) == nullptr) {
|
||||
return fail("unsupported parameter-band count");
|
||||
}
|
||||
if (input.n_dpoints[object] < 1u || input.n_dpoints[object] > 2u) {
|
||||
return fail("n_dpoints must be 1 or 2");
|
||||
}
|
||||
if (input.slope_idx[object] > 1u) {
|
||||
return fail("slope_idx must be 0 or 1");
|
||||
}
|
||||
if (input.slope_idx[object] == 1u) {
|
||||
const uint8_t* offsets =
|
||||
input.offset_ts + object * EJOC_MAX_DPOINTS;
|
||||
for (uint32_t point = 0; point < input.n_dpoints[object]; ++point) {
|
||||
if (offsets[point] >= EJOC_TIMESLOTS) {
|
||||
return fail("offset_ts leaves the 0..23 range");
|
||||
}
|
||||
if (point != 0u && offsets[point] <= offsets[point - 1u]) {
|
||||
return fail("offset_ts is not strictly increasing");
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if (n_channels_prev_ >= 0 && n_channels != static_cast<uint32_t>(n_channels_prev_)) {
|
||||
clear_unused_channels(n_channels);
|
||||
}
|
||||
// The parse status decides whether the frame's clipgain is usable; a
|
||||
// declared-channel-count change (status 9) is only the renderer's own
|
||||
// dispatch tag, so the new gain still applies to that frame.
|
||||
const uint32_t parse_status = status;
|
||||
// A declared core-channel-count change is the renderer's status 9: hard
|
||||
// cut with the LFE ring cleared, while the clipgain ramp state is kept.
|
||||
if (has_metadata && status == 0u && input.declared_core_channels != 0u &&
|
||||
n_channels_prev_ >= 0 &&
|
||||
input.declared_core_channels != static_cast<uint32_t>(n_channels_prev_)) {
|
||||
status = 9u;
|
||||
}
|
||||
advance_state(status, input.seq);
|
||||
maybe_clear_lfe_ring(status);
|
||||
const FrameMode mode = select_mode(status);
|
||||
if (input.phase_shift_valid != 0u) {
|
||||
phase_shift_enabled_ = input.phase_shift != 0u;
|
||||
}
|
||||
const bool valid_metadata =
|
||||
has_metadata && parse_status == 0u && input.declared_core_channels != 0u;
|
||||
double clipgain;
|
||||
float phase_new;
|
||||
if ((mode == FrameMode::Normal || mode == FrameMode::HardCut) && valid_metadata) {
|
||||
clipgain = input.clipgain;
|
||||
phase_new = static_cast<float>(0.0625 * clipgain);
|
||||
} else {
|
||||
// Held frames reuse the current clipgain; they do not retune it.
|
||||
clipgain = held_clipgain();
|
||||
phase_new = held_analysis_scalar();
|
||||
}
|
||||
|
||||
core_channels_ = n_channels;
|
||||
std::memset(input.output16_planar, 0,
|
||||
sizeof(float) * EJOC_OUTPUT_CHANNELS * EJOC_FRAME_SAMPLES);
|
||||
build_core8(input.bed_planar, input.lfe, n_channels);
|
||||
analysis(input.bed_planar, phase_new);
|
||||
|
||||
job_kind_ = JobKind::FrameObjects;
|
||||
frame_mode_ = mode;
|
||||
job_object_mask_ = mode == FrameMode::Normal ? input.object_mask : 0u;
|
||||
job_n_bands_ = input.n_bands;
|
||||
job_n_dpoints_ = input.n_dpoints;
|
||||
job_slope_idx_ = input.slope_idx;
|
||||
job_offset_ts_ = input.offset_ts;
|
||||
job_dq_ = input.dq;
|
||||
job_output_scale_ = input.output_scale;
|
||||
job_output16_ = input.output16_planar;
|
||||
job_count_ = EJOC_MAX_OBJECTS;
|
||||
for (int object = 0; object < EJOC_MAX_OBJECTS; ++object) {
|
||||
job_objects_[object] = object;
|
||||
}
|
||||
dispatch_jobs();
|
||||
|
||||
decode_lfe(input.lfe, clipgain, input.output16_planar, input.output_scale);
|
||||
if (mode == FrameMode::HardCut) {
|
||||
// All matrices are zero this frame; the presence markers stay set so
|
||||
// the next valid frame fades in from zero.
|
||||
std::memset(input.output16_planar, 0,
|
||||
sizeof(float) * EJOC_OUTPUT_CHANNELS * EJOC_FRAME_SAMPLES);
|
||||
for (int object = 0; object < EJOC_MAX_OBJECTS; ++object) {
|
||||
prev_present_[object] = true;
|
||||
for (int channel = 0; channel < EJOC_MAX_CORE_CHANNELS; ++channel) {
|
||||
std::memset(prev_matrix_[object][channel], 0,
|
||||
sizeof(prev_matrix_[object][channel]));
|
||||
}
|
||||
}
|
||||
} else if (mode == FrameMode::Passthrough) {
|
||||
write_passthrough(input.output16_planar, input.output_scale);
|
||||
}
|
||||
|
||||
// The 8-plane history is updated on every frame so a later passthrough
|
||||
// frame is continuous with the last core PCM.
|
||||
std::memcpy(core_history_, core8_, sizeof(core_history_));
|
||||
n_channels_prev_ = static_cast<int>(n_channels);
|
||||
return 0;
|
||||
}
|
||||
|
||||
enum class JobKind : uint8_t { Analysis, Objects, FrameObjects };
|
||||
|
||||
std::vector<std::thread> workers_;
|
||||
std::unique_ptr<std::barrier<>> work_barrier_;
|
||||
@@ -657,6 +1159,8 @@ private:
|
||||
const double* job_dq_ = nullptr;
|
||||
float job_output_scale_ = 1.0f;
|
||||
float* job_output16_ = nullptr;
|
||||
uint32_t job_object_mask_ = 0;
|
||||
FrameMode frame_mode_ = FrameMode::Normal;
|
||||
|
||||
uint32_t core_channels_ = EJOC_CORE_CHANNELS;
|
||||
alignas(64) double analysis_fifo_[EJOC_MAX_CORE_CHANNELS][9][64];
|
||||
@@ -668,11 +1172,29 @@ private:
|
||||
// sigma(k) switch; the caller sets it for downmix configurations 3 and 4.
|
||||
bool phase_shift_enabled_ = true;
|
||||
alignas(64) double lfe_delay_[kLfeDelay];
|
||||
// LFE gain ramp start; cleared by reset() and cleared/unset only there, so a
|
||||
// held frame keeps the last applied gain.
|
||||
double lfe_clipgain_ = 0.0;
|
||||
bool has_lfe_clipgain_ = false;
|
||||
alignas(64) double matrix_previous_[15][EJOC_MAX_CORE_CHANNELS][64];
|
||||
alignas(64) double synthesis_state_[15][640];
|
||||
alignas(64) Complex x_[EJOC_MAX_CORE_CHANNELS][64][24];
|
||||
alignas(64) Complex z_[15][64][24];
|
||||
|
||||
// Frame-level state: the seq state machine and the per-object interpolation
|
||||
// previous matrix (frame-end timeslot of the last published matrix).
|
||||
uint32_t state_ = 0;
|
||||
uint32_t expected_seq_ = 0;
|
||||
uint32_t invalid_count_ = 0;
|
||||
bool last_frame_valid_ = false;
|
||||
int n_channels_prev_ = -1;
|
||||
alignas(64) double prev_matrix_[EJOC_MAX_OBJECTS][EJOC_MAX_CORE_CHANNELS][64];
|
||||
bool prev_present_[EJOC_MAX_OBJECTS];
|
||||
// Passthrough history: the 8 core planes of the previous frame and of this
|
||||
// frame, in the order L,R,C,LFE,Ls,Rs,(Lrs,Rrs).
|
||||
alignas(64) double core_history_[EJOC_CORE_OUT_CHANNELS][EJOC_FRAME_SAMPLES];
|
||||
alignas(64) double core8_[EJOC_CORE_OUT_CHANNELS][EJOC_FRAME_SAMPLES];
|
||||
|
||||
uint8_t bit_reverse_[64];
|
||||
Complex fft_twiddle_[32];
|
||||
double rotation_sin_[64];
|
||||
@@ -690,13 +1212,13 @@ uint32_t EJOC_CALL ejoc_abi_version(void) {
|
||||
|
||||
const char* EJOC_CALL ejoc_build_info(void) {
|
||||
#if defined(_MSC_VER)
|
||||
return "eac3joc-core abi=1 compiler=MSVC fft=fixed64 speaker=double binaural=double crt=static-by-build";
|
||||
return "eac3joc-core abi=4 compiler=MSVC fft=fixed64 speaker=double binaural=double crt=static-by-build";
|
||||
#elif defined(__clang__)
|
||||
return "eac3joc-core abi=1 compiler=Clang fft=fixed64 speaker=double binaural=double";
|
||||
return "eac3joc-core abi=4 compiler=Clang fft=fixed64 speaker=double binaural=double";
|
||||
#elif defined(__GNUC__)
|
||||
return "eac3joc-core abi=1 compiler=GCC fft=fixed64 speaker=double binaural=double";
|
||||
return "eac3joc-core abi=4 compiler=GCC fft=fixed64 speaker=double binaural=double";
|
||||
#else
|
||||
return "eac3joc-core abi=1 compiler=unknown fft=fixed64 speaker=double binaural=double";
|
||||
return "eac3joc-core abi=4 compiler=unknown fft=fixed64 speaker=double binaural=double";
|
||||
#endif
|
||||
}
|
||||
|
||||
@@ -760,4 +1282,15 @@ int EJOC_CALL ejoc_renderer_process(
|
||||
slope_idx, offset_ts, dq, clipgain, phase_new, output_scale, output16_planar);
|
||||
}
|
||||
|
||||
int EJOC_CALL ejoc_renderer_render_frame(ejoc_renderer_handle handle,
|
||||
const ejoc_frame_input* input) {
|
||||
if (!handle) {
|
||||
return -1;
|
||||
}
|
||||
if (!input) {
|
||||
return -1;
|
||||
}
|
||||
return static_cast<ejoc::Renderer*>(handle)->render_frame(*input);
|
||||
}
|
||||
|
||||
} // extern "C"
|
||||
+38
-14
@@ -6,16 +6,17 @@
|
||||
namespace joc::joc {
|
||||
|
||||
Status rebuild_objects16(ejoc_renderer_handle handle, const joc_frame_params& params,
|
||||
const float* bed_planar, const float* lfe, float gain,
|
||||
const float* bed_planar, const float* lfe,
|
||||
std::uint32_t core_channel_count, bool has_metadata, float gain,
|
||||
std::vector<float>* out16, std::string* error) {
|
||||
if (handle == nullptr || bed_planar == nullptr || out16 == nullptr) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kDsp, "null argument");
|
||||
}
|
||||
if (params.n_channels != JOC_CORE_CHANNELS &&
|
||||
params.n_channels != JOC_MAX_CORE_CHANNELS) {
|
||||
if (core_channel_count != JOC_CORE_CHANNELS &&
|
||||
core_channel_count != JOC_MAX_CORE_CHANNELS) {
|
||||
if (error != nullptr) {
|
||||
*error = "the reused JOC kernel requires 5 or 7 core channels, frame declares " +
|
||||
std::to_string(static_cast<unsigned>(params.n_channels));
|
||||
*error = "the reused JOC kernel requires 5 or 7 core channels, the frame has " +
|
||||
std::to_string(core_channel_count);
|
||||
}
|
||||
return Status::fail(JOC_ERR_JOC_UNSUPPORTED_VARIANT, stage::kDsp, *error);
|
||||
}
|
||||
@@ -39,7 +40,7 @@ Status rebuild_objects16(ejoc_renderer_handle handle, const joc_frame_params& pa
|
||||
offset_ts[obj * JOC_MAX_DPOINTS + dp] = object.offset_ts[dp];
|
||||
}
|
||||
for (unsigned dp = 0; dp < object.n_dpoints; ++dp) {
|
||||
for (unsigned ch = 0; ch < params.n_channels; ++ch) {
|
||||
for (unsigned ch = 0; ch < core_channel_count; ++ch) {
|
||||
for (unsigned pb = 0; pb < object.n_bands; ++pb) {
|
||||
const std::size_t index =
|
||||
((static_cast<std::size_t>(obj) * JOC_MAX_DPOINTS + dp) *
|
||||
@@ -53,19 +54,42 @@ Status rebuild_objects16(ejoc_renderer_handle handle, const joc_frame_params& pa
|
||||
}
|
||||
}
|
||||
|
||||
out16->assign(static_cast<std::size_t>(JOC_OUTPUT_CHANNELS) * JOC_FRAME_SAMPLES, 0.0f);
|
||||
ejoc_frame_input input{};
|
||||
input.struct_size = sizeof(input);
|
||||
input.struct_version = EJOC_FRAME_INPUT_VERSION;
|
||||
input.bed_planar = bed_planar;
|
||||
input.lfe = lfe;
|
||||
input.core_channel_count = core_channel_count;
|
||||
// A frame without metadata has no declared channel count; the renderer's
|
||||
// status-9 path is gated on has_metadata, so the value is not consulted.
|
||||
input.declared_core_channels =
|
||||
has_metadata ? static_cast<std::uint32_t>(params.n_channels) : 0u;
|
||||
// sigma(k) = (k in {3,4}) gates the Ls/Rs -j rotation and the band-0 21-tap
|
||||
// DC compensation together; other configurations keep the common processing.
|
||||
const std::uint32_t phase_shift =
|
||||
(params.dmx_config_idx == 3 || params.dmx_config_idx == 4) ? 1u : 0u;
|
||||
const int result = ejoc_renderer_process(
|
||||
handle, bed_planar, lfe, params.n_channels, phase_shift, params.present_mask, n_bands,
|
||||
n_dpoints, slope_idx, offset_ts, dq.data(), params.clipgain, 0.0625f, gain,
|
||||
out16->data());
|
||||
// An unsupported downmix configuration (status 2) has no usable switch, so
|
||||
// the renderer keeps the previous one.
|
||||
input.phase_shift =
|
||||
(params.dmx_config_idx == 3u || params.dmx_config_idx == 4u) ? 1u : 0u;
|
||||
input.phase_shift_valid = params.status == JOC_FRAME_STATUS_UNSUPPORTED_DMX ? 0u : 1u;
|
||||
input.frame_status = has_metadata ? params.status : EJOC_FRAME_STATUS_NO_METADATA;
|
||||
input.seq = params.seq_count;
|
||||
input.has_metadata = has_metadata ? 1u : 0u;
|
||||
input.object_mask = params.present_mask;
|
||||
input.n_bands = n_bands;
|
||||
input.n_dpoints = n_dpoints;
|
||||
input.slope_idx = slope_idx;
|
||||
input.offset_ts = offset_ts;
|
||||
input.dq = dq.data();
|
||||
input.clipgain = params.clipgain;
|
||||
input.output_scale = gain;
|
||||
out16->assign(static_cast<std::size_t>(JOC_OUTPUT_CHANNELS) * JOC_FRAME_SAMPLES, 0.0f);
|
||||
input.output16_planar = out16->data();
|
||||
|
||||
const int result = ejoc_renderer_render_frame(handle, &input);
|
||||
if (result != 0) {
|
||||
const char* detail = ejoc_renderer_last_error(handle);
|
||||
const std::string message =
|
||||
"ejoc_renderer_process failed (" + std::to_string(result) + "): " +
|
||||
"ejoc_renderer_render_frame failed (" + std::to_string(result) + "): " +
|
||||
(detail != nullptr ? detail : "unknown");
|
||||
if (error != nullptr) {
|
||||
*error = message;
|
||||
|
||||
@@ -1,6 +1,7 @@
|
||||
|
||||
#pragma once
|
||||
|
||||
#include <cstdint>
|
||||
#include <string>
|
||||
#include <vector>
|
||||
|
||||
@@ -10,10 +11,20 @@
|
||||
|
||||
namespace joc::joc {
|
||||
|
||||
// [16][1536] planar float32. bed_planar is [params.n_channels][1536], where
|
||||
// n_channels comes from joc_dmx_config_idx (5 or 7).
|
||||
// Rebuilds the 16-channel object frame ([16][1536] planar float32) for one
|
||||
// audio frame through the renderer's frame-level state machine. bed_planar is
|
||||
// [core_channel_count][1536] in JOC lane order (L,R,C,Ls,Rs,(extra pair)), where
|
||||
// core_channel_count is the actual bed lane count (5 or 7) and lfe is the
|
||||
// separate core LFE plane (NULL = zeros; the LFE ring still advances).
|
||||
//
|
||||
// has_metadata is false for an audio frame without a JOC EMDF container: such a
|
||||
// frame always passes the 8 core planes through to output slots 0..7. For a
|
||||
// frame with metadata, params.status selects normal rendering or the invalid
|
||||
// frame path (hold / hard cut), and params.seq_count drives the state machine.
|
||||
// gain is the user output gain (not joc_clipgain).
|
||||
Status rebuild_objects16(ejoc_renderer_handle handle, const joc_frame_params& params,
|
||||
const float* bed_planar, const float* lfe, float gain,
|
||||
const float* bed_planar, const float* lfe,
|
||||
std::uint32_t core_channel_count, bool has_metadata, float gain,
|
||||
std::vector<float>* out16, std::string* error);
|
||||
|
||||
} // namespace joc::joc
|
||||
@@ -338,7 +338,7 @@ inline constexpr uint8_t kPbMap1[64] = {
|
||||
};
|
||||
|
||||
inline constexpr uint8_t kPbMap3[64] = {
|
||||
0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2,
|
||||
0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2,
|
||||
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
|
||||
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
|
||||
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
|
||||
@@ -346,8 +346,8 @@ inline constexpr uint8_t kPbMap3[64] = {
|
||||
|
||||
inline constexpr uint8_t kPbMap5[64] = {
|
||||
0, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3,
|
||||
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
|
||||
3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
|
||||
3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 4,
|
||||
4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
|
||||
4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4,
|
||||
};
|
||||
|
||||
@@ -359,8 +359,8 @@ inline constexpr uint8_t kPbMap7[64] = {
|
||||
};
|
||||
|
||||
inline constexpr uint8_t kPbMap9[64] = {
|
||||
0, 1, 2, 3, 3, 3, 4, 5, 5, 6, 6, 6, 7, 7, 8, 8,
|
||||
8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
|
||||
0, 1, 2, 3, 3, 4, 4, 5, 5, 6, 6, 6, 6, 6, 7, 7,
|
||||
7, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8, 8,
|
||||
8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
|
||||
8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8,
|
||||
};
|
||||
@@ -373,9 +373,9 @@ inline constexpr uint8_t kPbMap12[64] = {
|
||||
};
|
||||
|
||||
inline constexpr uint8_t kPbMap15[64] = {
|
||||
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 9, 10, 10, 11, 11, 11,
|
||||
12, 12, 12, 12, 13, 13, 13, 13, 13, 13, 14, 14, 14, 14, 14, 14,
|
||||
14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
|
||||
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 9, 10, 10, 10, 11, 11,
|
||||
11, 11, 12, 12, 12, 12, 12, 13, 13, 13, 13, 13, 13, 13, 13, 13,
|
||||
13, 13, 13, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
|
||||
14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14, 14,
|
||||
};
|
||||
|
||||
|
||||
File diff suppressed because it is too large.
Load diff
File diff suppressed because it is too large.
Load diff
@@ -0,0 +1,101 @@
|
||||
// Port of src/oamd_math.py: float32 equations for the metadata application path.
|
||||
//
|
||||
// The routines implement the fixed float32 equations for the size tables, ISF
|
||||
// basis assembly, bed channel routing and the position-dependent compensation.
|
||||
// The size tables are generated from the region geometry with grouped equal-power
|
||||
// panning, an S448 weight and a power sum; every operation is evaluated in
|
||||
// float32 in a fixed order so the coefficients are reproducible.
|
||||
//
|
||||
// The caller passes the parameter mappings from oamd_params; this module does
|
||||
// not read files itself.
|
||||
|
||||
#pragma once
|
||||
|
||||
#include <array>
|
||||
#include <cstddef>
|
||||
#include <cstdint>
|
||||
#include <vector>
|
||||
|
||||
#include "oamd/oamd_params.h"
|
||||
#include "speaker/speaker_layout_lookup.h"
|
||||
#include "speaker/speaker_layouts.h"
|
||||
|
||||
namespace joc::oamd {
|
||||
|
||||
// The specified 2**x approximation used by the compensation kernels.
|
||||
float exp2_approx(float value);
|
||||
|
||||
// The specified x**power approximation; power is a float32.
|
||||
float pow_approx(float value, float power);
|
||||
|
||||
// Stable float32 normalisation with the tiny-norm floor.
|
||||
std::vector<float> norm(const std::vector<float>& vector);
|
||||
|
||||
// One boundary weight of the size kernel.
|
||||
float boundary(int axis, int edge, float position, float t);
|
||||
|
||||
// Maps a scalar size in [0, 1] to t on the size curve.
|
||||
float size_curve_value(double size);
|
||||
|
||||
// Size kernel response in region-speaker order (the region point order).
|
||||
std::vector<float> size_kernel(const ::ejoc::speaker_tables::RegionGeometry& region,
|
||||
const float position[3], float t, bool height);
|
||||
|
||||
// Size-kernel target vector in layout channel order.
|
||||
std::vector<float> size_targets(const MetadataParams& meta,
|
||||
const ::ejoc::speaker_tables::LayoutGeometry& layout,
|
||||
int region_index, const double position[3],
|
||||
double size, bool height);
|
||||
|
||||
// One assembled ISF basis matrix (row_count x layout output_count, row major).
|
||||
struct IsfBasis {
|
||||
int rows = 0;
|
||||
int columns = 0;
|
||||
std::vector<float> values;
|
||||
|
||||
float at(int row, int column) const {
|
||||
return values[static_cast<std::size_t>(row) * static_cast<std::size_t>(columns) +
|
||||
static_cast<std::size_t>(column)];
|
||||
}
|
||||
};
|
||||
|
||||
// Assembles one ISF basis matrix from the compressed base tables.
|
||||
IsfBasis assemble_isf(const MetadataParams& meta, const char* layout_key, int isf_idx);
|
||||
|
||||
// Returns the per-basis ISF compensation gain.
|
||||
float isf_compensation(const MetadataParams& meta, const char* layout_key, int isf_idx,
|
||||
int row, bool disabled);
|
||||
|
||||
// Per-label bed redistribution weight, in the reference-position index space
|
||||
// used by the layout descriptors.
|
||||
float bed_weight(int label, std::uint32_t mask, int target_label);
|
||||
|
||||
// The five height-candidate flags for one input bed label.
|
||||
std::array<int, 5> bed_candidates(int label, std::uint32_t mask,
|
||||
const std::vector<int>& output_labels);
|
||||
|
||||
// Route and weights for one input bed label. route == -1 means the returned
|
||||
// weights are the target; otherwise route is the reference position used to
|
||||
// synthesise a point source.
|
||||
struct BedDistribution {
|
||||
int route = -1;
|
||||
std::vector<float> weights;
|
||||
};
|
||||
|
||||
// Derives the bed routing of one input bed label from the target layout.
|
||||
BedDistribution bed_distribution(const ::joc::speaker::LayoutInfo& layout, int label,
|
||||
std::uint32_t mask, bool distribute);
|
||||
|
||||
// Returns the Q15 reference position used by the bed fallback path.
|
||||
void bed_reference_position(int index, int q15[3]);
|
||||
|
||||
// Returns the bed path compensation gain for one input bed label. ``mode`` is
|
||||
// the trim mode (0 automatic, 1 disabled, 2 explicit) and ``levels`` holds the
|
||||
// surround, height and centre levels in dB.
|
||||
float bed_compensation(const ::joc::speaker::LayoutInfo& layout, int label, int mode,
|
||||
bool bypass, const double levels[3]);
|
||||
|
||||
// Automatic attenuation of a target layout in dB.
|
||||
float bed_attenuation_db(const ::joc::speaker::LayoutInfo& layout);
|
||||
|
||||
} // namespace joc::oamd
|
||||
@@ -0,0 +1,108 @@
|
||||
#include "oamd/oamd_params.h"
|
||||
|
||||
#include <cstddef>
|
||||
#include <utility>
|
||||
#include <vector>
|
||||
|
||||
#include "oamd/isf_tables.h"
|
||||
|
||||
namespace joc::oamd {
|
||||
|
||||
namespace {
|
||||
|
||||
namespace tables = isf_tables;
|
||||
|
||||
// Builds the renderer-facing view from the checked-in table header. There is no
|
||||
// file I/O and no runtime parsing: the numbers are compiled in as static data.
|
||||
MetadataParams build_params() {
|
||||
MetadataParams params;
|
||||
params.configs.reserve(tables::kConfigCount);
|
||||
for (const tables::Config& source : tables::kConfigs) {
|
||||
IsfConfig config;
|
||||
for (std::size_t index = 0; index < config.counts.size(); ++index) {
|
||||
config.counts[index] = source.counts[index];
|
||||
}
|
||||
const std::size_t rows = static_cast<std::size_t>(config.row_count());
|
||||
config.primary_row_map.assign(source.primary_row_map, source.primary_row_map + rows);
|
||||
config.secondary_row_map.assign(source.secondary_row_map,
|
||||
source.secondary_row_map + rows);
|
||||
const std::size_t coefficients = rows * static_cast<std::size_t>(tables::kMatrixColumns);
|
||||
config.matrix.assign(source.matrix, source.matrix + coefficients);
|
||||
params.configs.push_back(std::move(config));
|
||||
}
|
||||
|
||||
params.layouts.reserve(tables::kLayoutCount);
|
||||
for (const tables::Layout& source : tables::kLayouts) {
|
||||
IsfLayout layout;
|
||||
layout.output_count = source.output_count;
|
||||
for (std::size_t index = 0; index < layout.selectors.size(); ++index) {
|
||||
layout.selectors[index] = source.selectors[index];
|
||||
}
|
||||
layout.compensation_value = source.compensation_value;
|
||||
layout.output_ids.assign(source.output_ids,
|
||||
source.output_ids + tables::kOutputIdSlots);
|
||||
params.layouts.emplace(source.key, std::move(layout));
|
||||
}
|
||||
|
||||
const auto copy_groups = [](const tables::GroupRecord* records, int count,
|
||||
std::vector<IsfGroupRecord>* out) {
|
||||
out->reserve(static_cast<std::size_t>(count));
|
||||
for (int index = 0; index < count; ++index) {
|
||||
out->push_back(IsfGroupRecord{records[index].selector, records[index].begin,
|
||||
records[index].end});
|
||||
}
|
||||
};
|
||||
copy_groups(tables::kLayoutGroups.floor_a, tables::kLayoutGroups.floor_a_count,
|
||||
¶ms.layout_groups.floor_a);
|
||||
copy_groups(tables::kLayoutGroups.floor_b, tables::kLayoutGroups.floor_b_count,
|
||||
¶ms.layout_groups.floor_b);
|
||||
copy_groups(tables::kLayoutGroups.height, tables::kLayoutGroups.height_count,
|
||||
¶ms.layout_groups.height);
|
||||
copy_groups(tables::kLayoutGroups.additional, tables::kLayoutGroups.additional_count,
|
||||
¶ms.layout_groups.additional);
|
||||
|
||||
params.paired_columns.assign(tables::kPairedColumns,
|
||||
tables::kPairedColumns + tables::kPairedColumnCount);
|
||||
return params;
|
||||
}
|
||||
|
||||
const MetadataParams& loaded_params() {
|
||||
static const MetadataParams params = build_params();
|
||||
return params;
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
const std::vector<IsfGroupRecord>* IsfLayoutGroups::find(const char* name) const {
|
||||
if (name == nullptr) {
|
||||
return nullptr;
|
||||
}
|
||||
const std::string key(name);
|
||||
if (key == "floor_a") {
|
||||
return &floor_a;
|
||||
}
|
||||
if (key == "floor_b") {
|
||||
return &floor_b;
|
||||
}
|
||||
if (key == "height") {
|
||||
return &height;
|
||||
}
|
||||
if (key == "additional") {
|
||||
return &additional;
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
const IsfLayout* MetadataParams::layout(const char* key) const {
|
||||
if (key == nullptr) {
|
||||
return nullptr;
|
||||
}
|
||||
const auto found = layouts.find(key);
|
||||
return found != layouts.end() ? &found->second : nullptr;
|
||||
}
|
||||
|
||||
const MetadataParams* metadata_params() { return &loaded_params(); }
|
||||
|
||||
const char* metadata_params_error() { return ""; }
|
||||
|
||||
} // namespace joc::oamd
|
||||
@@ -0,0 +1,82 @@
|
||||
// Port of src/oamd_params.py: the static ISF lookup data of the metadata path.
|
||||
//
|
||||
// The numbers are compiled in as the generated header `oamd/isf_tables.h`:
|
||||
// ISF base coefficients and layout descriptors, applied as lookup tables.
|
||||
// The remaining values (gain, size, priority, position, delta and refinement
|
||||
// tables) are computed in code; the size tables are generated from the region
|
||||
// geometry at runtime, and bed routing is derived from the speaker layouts.
|
||||
// The arrays are adapted once per process into the structures below; nothing
|
||||
// reads a data file.
|
||||
|
||||
#pragma once
|
||||
|
||||
#include <array>
|
||||
#include <cstdint>
|
||||
#include <string>
|
||||
#include <unordered_map>
|
||||
#include <vector>
|
||||
|
||||
#include "foundation/status.h"
|
||||
|
||||
namespace joc::oamd {
|
||||
|
||||
// One ISF configuration: the compressed basis matrix and the two row maps
|
||||
// that expand it to the layout output channels.
|
||||
struct IsfConfig {
|
||||
// mid, upper, lower, zenith, row_count (the reference's counts tuple).
|
||||
std::array<int, 5> counts{};
|
||||
std::vector<int> primary_row_map;
|
||||
std::vector<int> secondary_row_map;
|
||||
// row_count * 73 float32 coefficients, row major.
|
||||
std::vector<float> matrix;
|
||||
|
||||
int row_count() const { return counts[4]; }
|
||||
};
|
||||
|
||||
// One layout descriptor: how many output channels the basis has,
|
||||
// the layout-group selectors and the automatic compensation value.
|
||||
struct IsfLayout {
|
||||
int output_count = 0;
|
||||
std::array<int, 4> selectors{}; // floor_a, floor_b, height, additional
|
||||
double compensation_value = 0.0;
|
||||
// ISF channel id of each matrix output column, padded with zeros to 37.
|
||||
std::vector<std::uint8_t> output_ids;
|
||||
};
|
||||
|
||||
// One record of a layout group: selector id and the half-open column range
|
||||
// [begin, end).
|
||||
struct IsfGroupRecord {
|
||||
int selector = 0;
|
||||
int begin = 0;
|
||||
int end = 0;
|
||||
};
|
||||
|
||||
// The four layout groups of the ISF tables.
|
||||
struct IsfLayoutGroups {
|
||||
std::vector<IsfGroupRecord> floor_a;
|
||||
std::vector<IsfGroupRecord> floor_b;
|
||||
std::vector<IsfGroupRecord> height;
|
||||
std::vector<IsfGroupRecord> additional;
|
||||
|
||||
const std::vector<IsfGroupRecord>* find(const char* name) const;
|
||||
};
|
||||
|
||||
// The ISF lookup data in the shape the metadata renderer consumes.
|
||||
struct MetadataParams {
|
||||
std::vector<IsfConfig> configs;
|
||||
std::unordered_map<std::string, IsfLayout> layouts;
|
||||
IsfLayoutGroups layout_groups;
|
||||
// One flag per matrix column: whether the column also contributes through
|
||||
// the secondary row map.
|
||||
std::vector<bool> paired_columns;
|
||||
|
||||
const IsfLayout* layout(const char* key) const;
|
||||
};
|
||||
|
||||
// Returns the process-wide tables. They are compiled in, so the pointer is
|
||||
// never null; ``metadata_params_error`` is kept for callers that log a load
|
||||
// failure and always returns an empty string.
|
||||
const MetadataParams* metadata_params();
|
||||
const char* metadata_params_error();
|
||||
|
||||
} // namespace joc::oamd
|
||||
+1057
-328
File diff suppressed because it is too large.
Load diff
+172
-11
@@ -9,6 +9,34 @@
|
||||
#include "foundation/status.h"
|
||||
#include "joc_core.h"
|
||||
|
||||
namespace joc::joc {
|
||||
|
||||
// OAMD Table 12 的标准 bed 声道。LSB 顺序与声明顺序一致:L/R、C、LFE、
|
||||
// Ls/Rs、Lrs/Rrs、TFL/TFR、TSL/TSR、TBL/TBR、LW/RW、LFE2。
|
||||
// The enum values are the reference's BED_LABEL_INDEX - 1.
|
||||
enum class BedChannel : std::uint8_t {
|
||||
kLeft = 0,
|
||||
kRight,
|
||||
kCenter,
|
||||
kLfe,
|
||||
kLeftSurround,
|
||||
kRightSurround,
|
||||
kLeftRearSurround,
|
||||
kRightRearSurround,
|
||||
kTopFrontLeft,
|
||||
kTopFrontRight,
|
||||
kTopSideLeft,
|
||||
kTopSideRight,
|
||||
kTopBackLeft,
|
||||
kTopBackRight,
|
||||
kLeftWide,
|
||||
kRightWide,
|
||||
kLfe2,
|
||||
kCount,
|
||||
};
|
||||
|
||||
} // namespace joc::joc
|
||||
|
||||
namespace joc::oamd {
|
||||
|
||||
inline constexpr int kMaxSlots = 16;
|
||||
@@ -17,8 +45,16 @@ inline constexpr int kAxes = 3;
|
||||
inline constexpr int kNQ12 = 62;
|
||||
inline constexpr int kNQ3 = 15;
|
||||
inline constexpr int kQ15Scale = 32768;
|
||||
// Largest ISF configuration of Table 11b (5.6.0.11). A frame cannot carry more
|
||||
// objects than this through any of the standard programmes, but the parser does
|
||||
// not reject a larger count: it only bounds the legacy 16-slot view.
|
||||
inline constexpr int kMaxObjects = 30;
|
||||
inline constexpr int kMaxBlocks = 8;
|
||||
inline constexpr std::uint32_t kTrimElementId = 2;
|
||||
inline constexpr std::uint32_t kObjectElementId = 1;
|
||||
inline constexpr std::uint32_t kExtendedElementId = 5;
|
||||
inline constexpr int kTrimConfigs = 9;
|
||||
inline constexpr int kBedLabelCount = 17;
|
||||
|
||||
// One of the nine per-configuration trim records of the OAMD trim element.
|
||||
// ``kAbsent`` means the configuration carries no data (the element only
|
||||
@@ -50,6 +86,92 @@ struct OamdTrim {
|
||||
bool disable_per_object[kMaxSlots] = {};
|
||||
};
|
||||
|
||||
// OAMD object kind: the programme assigns objects to the bed, to an ISF
|
||||
// configuration, or to the dynamic (positional) set, in that order.
|
||||
enum class ObjectKind : std::uint8_t {
|
||||
kBed = 0,
|
||||
kIsf = 1,
|
||||
kDynamic = 2,
|
||||
};
|
||||
|
||||
// Render state of one object in one block (5.5.9 object_info_block).
|
||||
struct ObjectRenderState {
|
||||
bool not_active = false;
|
||||
std::int32_t gain_q = 0; // integer gain, unity = 4096
|
||||
std::int32_t priority_q = 0; // Q10 priority, 32767 = default
|
||||
std::int32_t qx = 0; // Q15 room position, 0..32767
|
||||
std::int32_t qy = 0; // Q15 room position, 0..32767
|
||||
std::int32_t qz = 0; // signed Q15 elevation
|
||||
std::uint8_t zone_idx = 0;
|
||||
bool enable_elevation = true;
|
||||
std::uint8_t size_mode = 0; // 0 none, 1 isotropic, 2 per-axis
|
||||
std::uint16_t size_q[3] = {0, 0, 0}; // Q15 size per axis
|
||||
bool snap = false;
|
||||
std::uint16_t divergence_q = 0; // Q15 divergence width
|
||||
};
|
||||
|
||||
struct ObjectBlockState {
|
||||
ObjectRenderState render;
|
||||
};
|
||||
|
||||
// One object of the frame: its kind and the state of every block.
|
||||
struct ObjectElementState {
|
||||
ObjectKind kind = ObjectKind::kDynamic;
|
||||
std::vector<ObjectBlockState> blocks;
|
||||
};
|
||||
|
||||
// Timing shared by every object of one block. ``block_offset_samples`` already
|
||||
// includes the element's sample_offset, like the reference block dictionary.
|
||||
struct BlockTiming {
|
||||
std::uint32_t block_offset_samples = 0;
|
||||
std::uint32_t ramp_duration_samples = 0;
|
||||
};
|
||||
|
||||
// Program assignment (5.5.3) as consumed by the speaker renderer.
|
||||
struct ProgramState {
|
||||
bool dynamic_object_only = false;
|
||||
bool lfe_present = false;
|
||||
bool b_bed_chan_distribute = false;
|
||||
// program_assignment() content_description flags.
|
||||
bool content_reserved = false;
|
||||
bool content_bed = false;
|
||||
bool content_isf = false;
|
||||
bool content_dynamic = false;
|
||||
std::uint32_t num_bed_objects = 0;
|
||||
std::uint32_t num_isf_objects = 0;
|
||||
std::int32_t num_dynamic_objects = -1; // -1 when the frame declares none
|
||||
std::uint32_t bed_mask = 0; // 17-bit non-standard bed mask
|
||||
std::int32_t isf_idx = -1; // -1 when the frame declares none
|
||||
// Declared bed labels in channel order, including LFE/LFE2. Index i is the
|
||||
// label of object i for i < num_bed_objects.
|
||||
std::vector<::joc::joc::BedChannel> bed_labels;
|
||||
};
|
||||
|
||||
// One element whose parsed content ended past its declared size. The reference
|
||||
// keeps parsing such elements and only records the difference.
|
||||
struct ElementSizeMismatch {
|
||||
std::uint32_t element_ordinal = 0;
|
||||
std::uint32_t element_id = 0;
|
||||
std::uint32_t declared_size_bytes = 0;
|
||||
std::uint32_t declared_end_bit = 0;
|
||||
std::uint32_t parsed_end_bit = 0;
|
||||
};
|
||||
|
||||
// Full frame state (frame_state() in the reference): per-object per-block
|
||||
// render state plus programme assignment, timing and trim.
|
||||
struct FrameState {
|
||||
std::uint32_t object_count = 0;
|
||||
std::uint32_t block_count = 0;
|
||||
std::uint32_t sample_offset = 0; // object_element sample offset
|
||||
std::uint32_t ramp_duration_samples = 0; // block 0
|
||||
std::vector<BlockTiming> blocks;
|
||||
std::vector<ObjectElementState> objects;
|
||||
ProgramState program;
|
||||
bool has_trim = false;
|
||||
OamdTrim trim;
|
||||
std::vector<ElementSizeMismatch> size_mismatch;
|
||||
};
|
||||
|
||||
// (the reference's None), i.e. hold the previous position.
|
||||
struct SlotUpdate {
|
||||
std::int16_t q1 = -1;
|
||||
@@ -76,6 +198,12 @@ struct OamdUpdate {
|
||||
// keeps whatever controls the caller already holds.
|
||||
OamdTrim trim;
|
||||
bool has_trim = false;
|
||||
// Full per-object per-block state of this payload. ``has_frame_state`` is
|
||||
// true whenever the parser produced it; the speaker metadata path and the
|
||||
// binaural position timeline consume it, while the legacy slot fields above
|
||||
// stay authoritative for ADM export and the point-renderer path.
|
||||
bool has_frame_state = false;
|
||||
FrameState frame_state;
|
||||
};
|
||||
|
||||
class OamdState {
|
||||
@@ -103,6 +231,23 @@ public:
|
||||
|
||||
std::int16_t q(int slot, int axis) const { return q_[slot][axis]; }
|
||||
|
||||
// Resolves one frame-state position block for one object. The legacy
|
||||
// apply() entry above keeps the reference frame_update() view (16 slots,
|
||||
// block 0); the binaural position timeline uses this entry so a frame may
|
||||
// carry more objects and more blocks than that view can express. Callers
|
||||
// skip inactive and non-dynamic objects themselves, exactly like the
|
||||
// reference's frame_update() value construction.
|
||||
void apply_render_state(int object_index, const ObjectRenderState& render) {
|
||||
if (object_index < 0 || object_index >= kMaxObjects) {
|
||||
return;
|
||||
}
|
||||
q_[object_index][0] = static_cast<std::int16_t>(render.qx);
|
||||
q_[object_index][1] = static_cast<std::int16_t>(render.qy);
|
||||
const std::int32_t elevation = render.enable_elevation ? render.qz : 0;
|
||||
q_[object_index][2] =
|
||||
static_cast<std::int16_t>(elevation > 0 ? elevation : 0);
|
||||
}
|
||||
|
||||
// Objects 1..15 as q15 triples, the layout the speaker renderer consumes.
|
||||
void object_positions_q15(std::uint16_t out[kObjects][kAxes]) const {
|
||||
for (int object = 0; object < kObjects; ++object) {
|
||||
@@ -113,25 +258,18 @@ public:
|
||||
}
|
||||
|
||||
private:
|
||||
std::int16_t q_[kMaxSlots][kAxes] = {};
|
||||
std::int16_t q_[kMaxObjects][kAxes] = {};
|
||||
};
|
||||
|
||||
// q_of(k, n) = min(32767, floor(32768 k / n + 0.5)), the reference's quantiser.
|
||||
int q_of(int k, int n);
|
||||
|
||||
Status parse_id11(const std::uint8_t* payload, std::size_t payload_size, OamdUpdate* out);
|
||||
|
||||
// Parses only the trim element of a payload. ``out`` receives the first trim
|
||||
// element found; ``*present`` is false when the payload carries none. decode
|
||||
// paths that need the OAMD calibration but no object state can use this
|
||||
// without requiring an object element.
|
||||
Status parse_trim(const std::uint8_t* payload, std::size_t payload_size, OamdTrim* out,
|
||||
bool* present);
|
||||
|
||||
struct BedAssignment {
|
||||
bool lfe_only = false;
|
||||
bool standard = false;
|
||||
std::uint32_t mask = 0;
|
||||
std::uint32_t mask = 0; // 10-bit standard or 17-bit non-standard mask
|
||||
// Declared channels in channel order; LFE is kept in place.
|
||||
std::vector<::joc::joc::BedChannel> channels;
|
||||
};
|
||||
|
||||
struct ParseTrace {
|
||||
@@ -145,6 +283,26 @@ struct ParseTrace {
|
||||
bool reserved_data_not_present = false;
|
||||
};
|
||||
|
||||
// Legacy single-frame view (frame_update() in the reference): the 16-slot
|
||||
// position update plus programme and trim. A payload carrying more than 16
|
||||
// objects or more than one position block is rejected, exactly like the
|
||||
// reference legacy entry point. The ADM export path keeps this view; the
|
||||
// speaker and binaural renderers use parse_id11_state().
|
||||
Status parse_id11(const std::uint8_t* payload, std::size_t payload_size, OamdUpdate* out);
|
||||
|
||||
// Full frame_state() view: per-object per-block state, no 16-slot or
|
||||
// single-block restriction. The legacy fields are filled as well whenever the
|
||||
// frame fits them, so a caller can parse once and hand the same update to the
|
||||
// renderers and to the ADM export path where its shape allows it.
|
||||
Status parse_id11_state(const std::uint8_t* payload, std::size_t payload_size, OamdUpdate* out);
|
||||
|
||||
// Parses only the trim element of a payload. ``out`` receives the first trim
|
||||
// element found; ``*present`` is false when the payload carries none. decode
|
||||
// paths that need the OAMD calibration but no object state can use this
|
||||
// without requiring an object element.
|
||||
Status parse_trim(const std::uint8_t* payload, std::size_t payload_size, OamdTrim* out,
|
||||
bool* present);
|
||||
|
||||
Status parse_id11_verbose(const std::uint8_t* payload, std::size_t payload_size, OamdUpdate* out,
|
||||
ParseTrace* trace);
|
||||
|
||||
@@ -153,4 +311,7 @@ Status parse_id11_verbose(const std::uint8_t* payload, std::size_t payload_size,
|
||||
Status parse_id11_bed_assignments(const std::uint8_t* payload, std::size_t payload_size,
|
||||
std::vector<BedAssignment>* out);
|
||||
|
||||
// 10-bit standard bed mask -> 17-bit non-standard bed mask (5.6.1.1.4).
|
||||
std::uint32_t standard_bed_mask(std::uint32_t bed_mask);
|
||||
|
||||
} // namespace joc::oamd
|
||||
@@ -8,32 +8,55 @@ namespace {
|
||||
|
||||
struct FrozenLayout {
|
||||
const char* name;
|
||||
const char* compact_name;
|
||||
std::uint32_t out_ch_config;
|
||||
std::uint32_t speaker_bitfield;
|
||||
std::uint32_t surround_channels;
|
||||
std::uint32_t height_channels;
|
||||
bool carries_trim;
|
||||
const char* const* channel_names;
|
||||
};
|
||||
|
||||
const char* const kChannels20[] = {"L", "R"};
|
||||
const char* const kChannels31[] = {"L", "R", "C", "LFE"};
|
||||
const char* const kChannels40[] = {"L", "R", "Lrs", "Rrs"};
|
||||
const char* const kChannels51[] = {"L", "R", "C", "LFE", "Ls", "Rs"};
|
||||
const char* const kChannels71[] = {"L", "R", "C", "LFE", "Ls", "Rs", "Lrs", "Rrs"};
|
||||
const char* const kChannels512[] = {"L", "R", "C", "LFE", "Ls", "Rs", "Ltm", "Rtm"};
|
||||
const char* const kChannels514[] = {"L", "R", "C", "LFE", "Ls", "Rs", "Ltf", "Rtf",
|
||||
"Ltr", "Rtr"};
|
||||
const char* const kChannels712[] = {"L", "R", "C", "LFE", "Ls", "Rs", "Lrs", "Rrs",
|
||||
"Ltm", "Rtm"};
|
||||
const char* const kChannels714[] = {"L", "R", "C", "LFE", "Ls", "Rs", "Lrs", "Rrs",
|
||||
"Ltf", "Rtf", "Ltr", "Rtr"};
|
||||
const char* const kChannels914[] = {"L", "R", "C", "LFE", "Ls", "Rs", "Lrs", "Rrs",
|
||||
"Lw", "Rw", "Ltf", "Rtf", "Ltr", "Rtr"};
|
||||
const char* const kChannels916[] = {"L", "R", "C", "LFE", "Ls", "Rs", "Lrs", "Rrs",
|
||||
"Lw", "Rw", "Ltf", "Rtf", "Ltm", "Rtm", "Ltr", "Rtr"};
|
||||
const char* const kChannels60[] = {"L", "R", "Ls", "Rs", "Lrs", "Rrs"};
|
||||
|
||||
constexpr FrozenLayout kLayouts[kLayoutCount] = {
|
||||
{"2.0", 0, 1, 0, 0, true},
|
||||
{"3.1", 3, 7, 0, 0, true},
|
||||
{"4.0", 0xFF, 9, 2, 0, true},
|
||||
{"5.1", 7, 15, 2, 0, true},
|
||||
{"7.1", 11, 31, 4, 0, true},
|
||||
{"5.1.2", 13, 1039, 2, 2, true},
|
||||
{"5.1.4", 14, 2575, 2, 4, true},
|
||||
{"7.1.2", 15, 1055, 4, 2, false},
|
||||
{"7.1.4", 16, 2591, 4, 4, false},
|
||||
{"9.1.4", 19, 2719, 4, 4, false},
|
||||
{"9.1.6", 20, 3743, 4, 6, false},
|
||||
{"2.0", "20", 0, 1, 0, 0, true, kChannels20},
|
||||
{"3.1", "31", 3, 7, 0, 0, true, kChannels31},
|
||||
{"4.0", "40", 0xFF, 9, 2, 0, true, kChannels40},
|
||||
{"5.1", "51", 7, 15, 2, 0, true, kChannels51},
|
||||
{"7.1", "71", 11, 31, 4, 0, true, kChannels71},
|
||||
{"5.1.2", "512", 13, 1039, 2, 2, true, kChannels512},
|
||||
{"5.1.4", "514", 14, 2575, 2, 4, true, kChannels514},
|
||||
{"7.1.2", "712", 15, 1055, 4, 2, false, kChannels712},
|
||||
{"7.1.4", "714", 16, 2591, 4, 4, false, kChannels714},
|
||||
{"9.1.4", "914", 19, 2719, 4, 4, false, kChannels914},
|
||||
{"9.1.6", "916", 20, 3743, 4, 6, false, kChannels916},
|
||||
{"6.0", "60", 0xFF, 25, 4, 0, false, kChannels60},
|
||||
};
|
||||
|
||||
void fill(const FrozenLayout& source, LayoutInfo* out) {
|
||||
out->name = source.name;
|
||||
out->compact_name = source.compact_name;
|
||||
out->out_ch_config = source.out_ch_config;
|
||||
out->speaker_bitfield = source.speaker_bitfield;
|
||||
out->channel_count = ejoc_speaker_layout_channel_count(source.speaker_bitfield);
|
||||
out->channel_names = source.channel_names;
|
||||
out->surround_channels = source.surround_channels;
|
||||
out->height_channels = source.height_channels;
|
||||
out->carries_trim = source.carries_trim;
|
||||
|
||||
@@ -8,13 +8,18 @@
|
||||
|
||||
namespace joc::speaker {
|
||||
|
||||
inline constexpr int kLayoutCount = 11;
|
||||
inline constexpr int kLayoutCount = 12;
|
||||
|
||||
struct LayoutInfo {
|
||||
const char* name = "";
|
||||
// The reference's compact layout key ("2.0" -> "20", ...), the key of the
|
||||
// automatic-attenuation and divide-by-default tables.
|
||||
const char* compact_name = "";
|
||||
std::uint32_t out_ch_config = 0;
|
||||
std::uint32_t speaker_bitfield = 0;
|
||||
std::uint32_t channel_count = 0;
|
||||
// Standard channel names in output order, ``channel_count`` entries.
|
||||
const char* const* channel_names = nullptr;
|
||||
// Ls/Rs/Lrs/Rrs and Ltf/Rtf/Ltr/Rtr/Ltm/Rtm, the labels the OAMD trim
|
||||
// configuration index counts.
|
||||
std::uint32_t surround_channels = 0;
|
||||
@@ -24,8 +29,8 @@ struct LayoutInfo {
|
||||
bool carries_trim = false;
|
||||
};
|
||||
|
||||
// All eleven layouts in the reference's order (2.0, 3.1, 4.0, 5.1, 7.1, 5.1.2,
|
||||
// 5.1.4, 7.1.2, 7.1.4, 9.1.4, 9.1.6).
|
||||
// All twelve layouts in the reference's order (2.0, 3.1, 4.0, 5.1, 7.1, 5.1.2,
|
||||
// 5.1.4, 7.1.2, 7.1.4, 9.1.4, 9.1.6, 6.0).
|
||||
bool layout_at(int index, LayoutInfo* out);
|
||||
|
||||
// False when the name is unknown (case-sensitive, as the reference CLI). "4.0"
|
||||
|
||||
@@ -39,7 +39,7 @@ struct LayoutGeometry {
|
||||
std::array<RegionGeometry, 7> regions{};
|
||||
};
|
||||
|
||||
inline constexpr std::array<LayoutGeometry, 11> kLayouts{{
|
||||
inline constexpr std::array<LayoutGeometry, 12> kLayouts{{
|
||||
LayoutGeometry{
|
||||
0x1u, 0, 2,
|
||||
{{0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}},
|
||||
@@ -159,9 +159,9 @@ inline constexpr std::array<LayoutGeometry, 11> kLayouts{{
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
2, 1, 1, 0,
|
||||
{{SpeakerPoint{{0, 32767, 0}, 2}, SpeakerPoint{{32767, 32767, 0}, 3}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}},
|
||||
4, 2, 2, 0,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{0, 32767, 0}, 2}, SpeakerPoint{{32767, 32767, 0}, 3}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{2, {2, 3, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
@@ -190,25 +190,25 @@ inline constexpr std::array<LayoutGeometry, 11> kLayouts{{
|
||||
{{
|
||||
RegionGeometry{
|
||||
5, 2, 2, 0,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{3, {0, 2, 1}}, AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
5, 2, 2, 0,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{3, {0, 2, 1}}, AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
5, 2, 2, 0,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{3, {0, 2, 1}}, AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
3, 2, 2, 0,
|
||||
{{SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{1, {0, 0, 0}}, AxisGroup{2, {1, 2, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
@@ -220,7 +220,7 @@ inline constexpr std::array<LayoutGeometry, 11> kLayouts{{
|
||||
},
|
||||
RegionGeometry{
|
||||
2, 1, 1, 0,
|
||||
{{SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
@@ -286,25 +286,25 @@ inline constexpr std::array<LayoutGeometry, 11> kLayouts{{
|
||||
{{
|
||||
RegionGeometry{
|
||||
7, 3, 2, 1,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{3, {0, 2, 1}}, AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {5, 6, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
7, 3, 2, 1,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{3, {0, 2, 1}}, AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {5, 6, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
7, 3, 2, 1,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{3, {0, 2, 1}}, AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {5, 6, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
5, 3, 2, 1,
|
||||
{{SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{1, {0, 0, 0}}, AxisGroup{2, {1, 2, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
@@ -316,7 +316,7 @@ inline constexpr std::array<LayoutGeometry, 11> kLayouts{{
|
||||
},
|
||||
RegionGeometry{
|
||||
4, 3, 1, 1,
|
||||
{{SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 16384, 32767}, 6}, SpeakerPoint{{24840, 16384, 32767}, 7}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {2, 3, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
@@ -334,25 +334,25 @@ inline constexpr std::array<LayoutGeometry, 11> kLayouts{{
|
||||
{{
|
||||
RegionGeometry{
|
||||
9, 3, 2, 2,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{3, {0, 2, 1}}, AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {5, 6, 0}}, AxisGroup{2, {7, 8, 0}}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
9, 3, 2, 2,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{3, {0, 2, 1}}, AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {5, 6, 0}}, AxisGroup{2, {7, 8, 0}}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
9, 3, 2, 2,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{3, {0, 2, 1}}, AxisGroup{2, {3, 4, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {5, 6, 0}}, AxisGroup{2, {7, 8, 0}}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
7, 3, 2, 2,
|
||||
{{SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{16384, 0, 0}, 2}, SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{1, {0, 0, 0}}, AxisGroup{2, {1, 2, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {3, 4, 0}}, AxisGroup{2, {5, 6, 0}}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
@@ -364,7 +364,7 @@ inline constexpr std::array<LayoutGeometry, 11> kLayouts{{
|
||||
},
|
||||
RegionGeometry{
|
||||
6, 3, 1, 2,
|
||||
{{SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{SpeakerPoint{{0, 16384, 0}, 4}, SpeakerPoint{{32767, 16384, 0}, 5}, SpeakerPoint{{7928, 7928, 32767}, 6}, SpeakerPoint{{24840, 7928, 32767}, 7}, SpeakerPoint{{7928, 24840, 32767}, 8}, SpeakerPoint{{24840, 24840, 32767}, 9}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{2, {2, 3, 0}}, AxisGroup{2, {4, 5, 0}}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
@@ -567,7 +567,55 @@ inline constexpr std::array<LayoutGeometry, 11> kLayouts{{
|
||||
{{AxisGroup{2, {5, 6, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
}
|
||||
}}
|
||||
},
|
||||
LayoutGeometry{
|
||||
0x19u, 0xFFu, 6,
|
||||
{{0, 1, 2, 3, 4, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}},
|
||||
{{
|
||||
RegionGeometry{
|
||||
6, 2, 3, 0,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{0, 16384, 0}, 2}, SpeakerPoint{{32767, 16384, 0}, 3}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{2, {2, 3, 0}}, AxisGroup{2, {4, 5, 0}}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
4, 2, 2, 0,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{0, 16384, 0}, 2}, SpeakerPoint{{32767, 16384, 0}, 3}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{2, {2, 3, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
4, 2, 2, 0,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{2, {2, 3, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
2, 2, 1, 0,
|
||||
{{SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
2, 1, 1, 0,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
4, 2, 2, 0,
|
||||
{{SpeakerPoint{{0, 16384, 0}, 2}, SpeakerPoint{{32767, 16384, 0}, 3}, SpeakerPoint{{0, 32767, 0}, 4}, SpeakerPoint{{32767, 32767, 0}, 5}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{2, {2, 3, 0}}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
},
|
||||
RegionGeometry{
|
||||
2, 1, 1, 0,
|
||||
{{SpeakerPoint{{0, 0, 0}, 0}, SpeakerPoint{{32767, 0, 0}, 1}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}, SpeakerPoint{}}},
|
||||
{{AxisGroup{2, {0, 1, 0}}, AxisGroup{}, AxisGroup{}, AxisGroup{}}},
|
||||
{{AxisGroup{}, AxisGroup{}, AxisGroup{}, AxisGroup{}}}
|
||||
}
|
||||
}}
|
||||
},
|
||||
}};
|
||||
|
||||
inline constexpr const LayoutGeometry* find_layout(const std::uint32_t speaker_bitfield) noexcept {
|
||||
|
||||
@@ -0,0 +1,90 @@
|
||||
// Port of the metadata application path of src/speaker_renderer.py HEAD.
|
||||
//
|
||||
// The metadata path consumes the full per-object per-block OAMD frame state and
|
||||
// schedules block targets frame by frame: object gain/inactive, zone/elevation,
|
||||
// snap, divergence, size kernel fusion, priority size budget, ISF basis rows and
|
||||
// bed channel distribution, with the same float64 gain ramps and block alignment
|
||||
// as the reference renderer.
|
||||
|
||||
#pragma once
|
||||
|
||||
#include <cstdint>
|
||||
#include <memory>
|
||||
#include <vector>
|
||||
|
||||
#include "eac3joc_core.h"
|
||||
#include "foundation/status.h"
|
||||
#include "oamd/oamd_parser.h"
|
||||
#include "speaker/speaker_layout_lookup.h"
|
||||
#include "speaker/speaker_trim.h"
|
||||
|
||||
namespace joc::speaker {
|
||||
|
||||
// One metadata target matrix: object_count x layout channel_count in the
|
||||
// renderer's internal channel order, before the ramp/timing stage.
|
||||
struct MetadataTargets {
|
||||
std::uint32_t object_count = 0;
|
||||
std::uint32_t channel_count = 0;
|
||||
std::vector<double> values; // object major
|
||||
|
||||
double at(std::uint32_t object, std::uint32_t channel) const {
|
||||
return values[static_cast<std::size_t>(object) * channel_count + channel];
|
||||
}
|
||||
};
|
||||
|
||||
// Computes the metadata target matrix of one block of a parsed frame without
|
||||
// touching renderer state: the reference's _init_meta validation plus
|
||||
// _block_targets. ``trim`` carries the current programme trim (bypass flags),
|
||||
// ``controls`` the derived per-layout controls, ``size_budget`` the priority
|
||||
// budget (0 means unlimited, the production value).
|
||||
Status metadata_block_targets(const LayoutInfo& layout, const oamd::OamdUpdate& update,
|
||||
std::uint32_t block_index,
|
||||
const ejoc_speaker_controls& controls,
|
||||
const oamd::OamdTrim* trim, std::uint32_t size_budget,
|
||||
MetadataTargets* out);
|
||||
|
||||
// Stateful metadata-path renderer. It owns the object target ramps and the
|
||||
// cross-block pending schedule; the legacy point-renderer state stays in the
|
||||
// C ABI renderer, which is used for the blocks of a frame that precede the first
|
||||
// metadata targets (the reference renders those with its legacy block path).
|
||||
class MetaRenderer {
|
||||
public:
|
||||
MetaRenderer();
|
||||
~MetaRenderer();
|
||||
MetaRenderer(MetaRenderer&&) noexcept;
|
||||
MetaRenderer& operator=(MetaRenderer&&) noexcept;
|
||||
MetaRenderer(const MetaRenderer&) = delete;
|
||||
MetaRenderer& operator=(const MetaRenderer&) = delete;
|
||||
|
||||
// Binds the target layout. Must be called once before process_frame().
|
||||
Status configure(const LayoutInfo& layout);
|
||||
|
||||
// Drops the ramp, schedule and metadata state (the layout stays bound).
|
||||
Status reset();
|
||||
|
||||
bool configured() const;
|
||||
// True once a metadata frame has been accepted (the reference's _meta_mode).
|
||||
bool engaged() const;
|
||||
|
||||
const char* last_error() const;
|
||||
|
||||
Status process_frame(ejoc_speaker_renderer_handle legacy,
|
||||
const float* objects16_interleaved, std::uint32_t sample_count,
|
||||
const oamd::OamdUpdate* update, std::uint32_t metadata_offset,
|
||||
const ejoc_speaker_controls& controls, const oamd::OamdTrim* trim,
|
||||
std::uint32_t raw_trim_mode, std::uint32_t size_budget,
|
||||
double* output_interleaved);
|
||||
|
||||
// Computes the target matrix of one block of ``update`` into ``out`` using a
|
||||
// scratch context; unlike metadata_block_targets() this validates through
|
||||
// the renderer's own state-independent path.
|
||||
Status block_targets(const oamd::OamdUpdate& update, std::uint32_t block_index,
|
||||
const ejoc_speaker_controls& controls, const oamd::OamdTrim* trim,
|
||||
std::uint32_t size_budget, MetadataTargets* out) const;
|
||||
|
||||
private:
|
||||
struct Impl;
|
||||
std::unique_ptr<Impl> impl_;
|
||||
};
|
||||
|
||||
} // namespace joc::speaker
|
||||
File diff suppressed because it is too large.
Load diff
@@ -37,18 +37,52 @@ Status step(SpeakerStep* context, const std::vector<float>& objects16_planar,
|
||||
}
|
||||
context->controls_ready = true;
|
||||
}
|
||||
std::uint32_t ramp = 0;
|
||||
if (update != nullptr) {
|
||||
if (update->has_trim) {
|
||||
const Status status = parameters_from_oamd(&update->trim, context->layout,
|
||||
&context->controls);
|
||||
if (update != nullptr && update->has_trim) {
|
||||
const Status status =
|
||||
parameters_from_oamd(&update->trim, context->layout, &context->controls);
|
||||
if (!status.ok()) {
|
||||
if (error != nullptr) {
|
||||
*error = status.message();
|
||||
}
|
||||
return status;
|
||||
}
|
||||
context->trim = update->trim;
|
||||
context->has_trim = true;
|
||||
context->raw_trim_mode = update->trim.global_trim_mode;
|
||||
}
|
||||
|
||||
const bool metadata_frame =
|
||||
update != nullptr && update->has_frame_state &&
|
||||
update->frame_state.object_count != 0u;
|
||||
if (metadata_frame || context->meta_engaged) {
|
||||
if (!context->meta_configured) {
|
||||
const Status configured = context->meta.configure(context->layout);
|
||||
if (!configured.ok()) {
|
||||
if (error != nullptr) {
|
||||
*error = configured.message();
|
||||
}
|
||||
return configured;
|
||||
}
|
||||
context->meta_configured = true;
|
||||
}
|
||||
const oamd::OamdTrim* trim = context->has_trim ? &context->trim : nullptr;
|
||||
const Status processed = context->meta.process_frame(
|
||||
context->handle, context->interleaved.data(), JOC_FRAME_SAMPLES,
|
||||
metadata_frame ? update : nullptr, context->metadata_offset,
|
||||
context->controls, trim, context->raw_trim_mode, 0u,
|
||||
context->output.data());
|
||||
if (!processed.ok()) {
|
||||
if (error != nullptr) {
|
||||
*error = processed.message();
|
||||
}
|
||||
return processed;
|
||||
}
|
||||
context->meta_engaged = context->meta.engaged();
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
std::uint32_t ramp = 0;
|
||||
if (update != nullptr) {
|
||||
context->state.apply(*update);
|
||||
ramp = update->ramp_duration_samples;
|
||||
context->last_block_offset = update->block_offset_samples;
|
||||
|
||||
@@ -9,6 +9,7 @@
|
||||
#include "foundation/status.h"
|
||||
#include "oamd/oamd_parser.h"
|
||||
#include "speaker/speaker_layout_lookup.h"
|
||||
#include "speaker/speaker_meta.h"
|
||||
#include "speaker/speaker_trim.h"
|
||||
|
||||
namespace joc::speaker {
|
||||
@@ -32,6 +33,16 @@ struct SpeakerStep {
|
||||
// 对象槽 → Q15 方位;bed_override 为 false 的槽仍按 OAMD 坐标声像。
|
||||
std::uint16_t bed_positions[oamd::kObjects][oamd::kAxes] = {};
|
||||
bool bed_override[oamd::kObjects] = {};
|
||||
// Metadata path: a frame that carries the full per-object block state
|
||||
// switches the speaker output to the metadata renderer. The programme trim
|
||||
// element and its raw global mode stay here because the metadata path needs
|
||||
// the per-object bypass flags and the ISF compensation mode.
|
||||
MetaRenderer meta;
|
||||
bool meta_configured = false;
|
||||
bool meta_engaged = false;
|
||||
oamd::OamdTrim trim{};
|
||||
bool has_trim = false;
|
||||
std::uint32_t raw_trim_mode = 0;
|
||||
};
|
||||
|
||||
Status step(SpeakerStep* context, const std::vector<float>& objects16_planar,
|
||||
|
||||
+81
-8
@@ -6,6 +6,7 @@
|
||||
|
||||
#include "adm/adm_metadata.h"
|
||||
#include "foundation/status.h"
|
||||
#include "virtual_speaker/virtual_speaker.h"
|
||||
#include "hrtf/jochrtf.h"
|
||||
#include "hrtf/rosella_model.h"
|
||||
#include "joc_bitstream/bed_layout_map.h"
|
||||
@@ -45,6 +46,7 @@ void Stream::reset_state() {
|
||||
bed_pending_.clear();
|
||||
bed_read_offset_ = 0;
|
||||
objects16_.clear();
|
||||
pcm_input_.reset();
|
||||
objects_pending_.clear();
|
||||
objects_read_offset_ = 0;
|
||||
output_.clear();
|
||||
@@ -77,7 +79,7 @@ Status Stream::create(const Config& config) {
|
||||
return Status::fail(JOC_ERR_STATE, stage::kRender, "stream already created");
|
||||
}
|
||||
config_ = config;
|
||||
if (config_.input > JOC_STREAM_IN_PCM_OBJECTS16 ||
|
||||
if (config_.input > JOC_STREAM_IN_PCM_LAYOUT ||
|
||||
config_.output > JOC_STREAM_OUT_BINAURAL) {
|
||||
return Status::fail(JOC_ERR_INVALID_CONFIG, stage::kRender, "unknown stream kind");
|
||||
}
|
||||
@@ -87,6 +89,20 @@ Status Stream::create(const Config& config) {
|
||||
"objects16 input with objects16 output would do nothing");
|
||||
}
|
||||
gain_ = static_cast<float>(std::pow(10.0, config_.gain_db / 20.0));
|
||||
if (config_.input == JOC_STREAM_IN_PCM_LAYOUT && !config_.pcm_layout.empty() &&
|
||||
config_.pcm_layout != "auto") {
|
||||
const virtual_speaker::Layout* layout =
|
||||
virtual_speaker::find_layout(config_.pcm_layout);
|
||||
if (layout == nullptr) {
|
||||
return Status::fail(JOC_ERR_INVALID_CONFIG, stage::kRender,
|
||||
"unknown PCM layout: " + config_.pcm_layout);
|
||||
}
|
||||
const Status configured =
|
||||
pcm_input_.configure(config_.pcm_layout, layout->channels, config_.gain_db);
|
||||
if (!configured.ok()) {
|
||||
return configured;
|
||||
}
|
||||
}
|
||||
|
||||
if (config_.input == JOC_STREAM_IN_EAC3) {
|
||||
rebuilder_ = ejoc_renderer_create();
|
||||
@@ -246,6 +262,9 @@ Status Stream::flush_pending_frame() {
|
||||
if (entry.params.objects_clamped != 0u) {
|
||||
++info_.objects_clamped_frames;
|
||||
}
|
||||
if (entry.params.status == JOC_FRAME_STATUS_TRUNCATED) {
|
||||
++info_.truncated_joc_frames;
|
||||
}
|
||||
} else {
|
||||
// No JOC metadata for this audio frame: the object channels stay silent
|
||||
// while the kernel state still advances for the frame.
|
||||
@@ -343,7 +362,14 @@ Status Stream::push_eac3(const std::uint8_t* data, std::size_t size, std::size_t
|
||||
if (!extracted.ok()) {
|
||||
return extracted;
|
||||
}
|
||||
const Status oamd = oamd::parse_id11(bytes.data(), bytes.size(), &pending_frame_.update);
|
||||
// Speaker and binaural consume the full per-object block state; the
|
||||
// other stream paths keep the reference's legacy single-block view.
|
||||
const bool full_metadata_state = speaker_enabled_ || binaural_enabled_;
|
||||
const Status oamd =
|
||||
full_metadata_state ? oamd::parse_id11_state(bytes.data(), bytes.size(),
|
||||
&pending_frame_.update)
|
||||
: oamd::parse_id11(bytes.data(), bytes.size(),
|
||||
&pending_frame_.update);
|
||||
if (!oamd.ok()) {
|
||||
return oamd;
|
||||
}
|
||||
@@ -443,6 +469,47 @@ Status Stream::push_objects16(const float* planar16, std::size_t samples, std::s
|
||||
return process_objects16_frames(false);
|
||||
}
|
||||
|
||||
Status Stream::push_pcm_layout(const float* interleaved, std::size_t samples,
|
||||
std::uint32_t channels, std::size_t* consumed) {
|
||||
if (consumed != nullptr) {
|
||||
*consumed = 0;
|
||||
}
|
||||
if (interleaved == nullptr || samples == 0u) {
|
||||
return Status::success();
|
||||
}
|
||||
if (pcm_input_.layout() == nullptr) {
|
||||
const Status configured = pcm_input_.configure(config_.pcm_layout,
|
||||
static_cast<int>(channels),
|
||||
config_.gain_db);
|
||||
if (!configured.ok()) {
|
||||
return configured;
|
||||
}
|
||||
} else if (pcm_input_.channels() != static_cast<int>(channels)) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kRender,
|
||||
"PCM channel count changed mid-stream");
|
||||
}
|
||||
Status status = pcm_input_.append(interleaved, samples);
|
||||
if (!status.ok()) {
|
||||
return status;
|
||||
}
|
||||
std::size_t processed = 0;
|
||||
while (pcm_input_.frame_ready()) {
|
||||
status = pcm_input_.take_frame(&objects_frame_);
|
||||
if (!status.ok()) {
|
||||
return status;
|
||||
}
|
||||
status = push_objects16(objects_frame_.data(), kFrameSamples, nullptr);
|
||||
if (!status.ok()) {
|
||||
return status;
|
||||
}
|
||||
processed += kFrameSamples;
|
||||
}
|
||||
if (consumed != nullptr) {
|
||||
*consumed = processed;
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
Status Stream::process_objects16_frames(bool drain_all) {
|
||||
constexpr std::size_t kFrameValues =
|
||||
static_cast<std::size_t>(JOC_OUTPUT_CHANNELS) * kFrameSamples;
|
||||
@@ -536,8 +603,10 @@ Status Stream::process_ready_frames(bool drain_all) {
|
||||
compact_bed_pending();
|
||||
|
||||
std::string error;
|
||||
const Status rebuilt = joc::rebuild_objects16(rebuilder_, entry.params, bed5_.data(),
|
||||
lfe_.data(), gain_, &objects16_, &error);
|
||||
const Status rebuilt = joc::rebuild_objects16(
|
||||
rebuilder_, entry.params, bed5_.data(), lfe_.data(),
|
||||
static_cast<std::uint32_t>(core_channels), entry.has_joc, gain_, &objects16_,
|
||||
&error);
|
||||
if (!rebuilt.ok()) {
|
||||
return Status::fail(rebuilt.code(), stage::kDsp, error);
|
||||
}
|
||||
@@ -578,9 +647,11 @@ Status Stream::render_objects16(const std::vector<float>& objects16) {
|
||||
if (speaker_enabled_) {
|
||||
std::string error;
|
||||
const Status stepped =
|
||||
speaker::step(&speaker_, objects16, pending_metadata_.has_update
|
||||
speaker::step(&speaker_, objects16,
|
||||
pending_metadata_.has_update
|
||||
? &pending_metadata_.update
|
||||
: nullptr,
|
||||
: (pcm_input_.layout() != nullptr ? &pcm_input_.fixed_update()
|
||||
: nullptr),
|
||||
&error);
|
||||
if (!stepped.ok()) {
|
||||
return Status::fail(stepped.code(), stage::kRender, error);
|
||||
@@ -599,7 +670,8 @@ Status Stream::render_objects16(const std::vector<float>& objects16) {
|
||||
|
||||
const Status submitted =
|
||||
binaural_.submit_frame(objects16.data(),
|
||||
pending_metadata_.has_update ? &pending_metadata_.update : nullptr,
|
||||
pending_metadata_.has_update ? &pending_metadata_.update
|
||||
: (pcm_input_.layout() != nullptr ? &pcm_input_.fixed_update() : nullptr),
|
||||
static_cast<std::int64_t>(info_.frames_out),
|
||||
pending_metadata_.outer_offset,
|
||||
static_cast<std::int64_t>(config_.object_delay_samples));
|
||||
@@ -626,7 +698,8 @@ Status Stream::render_objects16(const std::vector<float>& objects16) {
|
||||
Status Stream::render_rosella_objects16(const std::vector<float>& objects16) {
|
||||
const Status submitted =
|
||||
rosella_.submit_frame(objects16.data(),
|
||||
pending_metadata_.has_update ? &pending_metadata_.update : nullptr,
|
||||
pending_metadata_.has_update ? &pending_metadata_.update
|
||||
: (pcm_input_.layout() != nullptr ? &pcm_input_.fixed_update() : nullptr),
|
||||
static_cast<std::int64_t>(info_.frames_out),
|
||||
pending_metadata_.outer_offset,
|
||||
static_cast<std::int64_t>(config_.object_delay_samples));
|
||||
|
||||
@@ -15,6 +15,7 @@
|
||||
#include "joc_stream.h"
|
||||
#include "oamd/oamd_parser.h"
|
||||
#include "speaker/speaker_step.h"
|
||||
#include "virtual_speaker/virtual_speaker.h"
|
||||
|
||||
namespace joc::stream {
|
||||
|
||||
@@ -34,6 +35,7 @@ struct Config {
|
||||
// (.jochrtf).
|
||||
std::string personalized_headphone_path;
|
||||
std::uint32_t jochrtf_room = 1;
|
||||
std::string pcm_layout;
|
||||
// Filterbank of the Rosella path; the other binaural inputs ignore it.
|
||||
std::uint32_t rosella_filterbank = JOC_ROSELLA_FILTERBANK_PUBLIC;
|
||||
};
|
||||
@@ -47,6 +49,9 @@ struct Info {
|
||||
std::uint64_t oamd_payloads = 0;
|
||||
std::uint64_t oamd_transitions = 0;
|
||||
std::uint64_t missing_joc_frames = 0;
|
||||
// Audio frames whose ID14 payload ended inside joc_data; they render as
|
||||
// invalid frames and hold the previous matrices.
|
||||
std::uint64_t truncated_joc_frames = 0;
|
||||
std::uint64_t objects_clamped_frames = 0;
|
||||
// Audio frames whose syncframe held more than one top-level JOC container and
|
||||
// were decoded from the single one that carries a complete joc() payload.
|
||||
@@ -96,6 +101,8 @@ public:
|
||||
Status push_bed(const float* interleaved, std::size_t samples, std::uint32_t channels,
|
||||
std::size_t* consumed);
|
||||
Status push_objects16(const float* planar16, std::size_t samples, std::size_t* consumed);
|
||||
Status push_pcm_layout(const float* interleaved, std::size_t samples, std::uint32_t channels,
|
||||
std::size_t* consumed);
|
||||
Status pull(float* destination, std::size_t capacity_samples, std::size_t* produced);
|
||||
Status flush();
|
||||
Status reset();
|
||||
@@ -157,6 +164,9 @@ private:
|
||||
std::size_t bed_read_offset_ = 0;
|
||||
std::vector<float> bed5_;
|
||||
std::vector<float> lfe_;
|
||||
virtual_speaker::Input pcm_input_;
|
||||
oamd::OamdUpdate fixed_update_{};
|
||||
bool has_fixed_update_ = false;
|
||||
std::vector<float> objects_pending_;
|
||||
std::size_t objects_read_offset_ = 0;
|
||||
std::vector<float> objects_frame_;
|
||||
|
||||
+63
-6
@@ -348,7 +348,7 @@ Status probe_atmos_trim(const std::string& path, std::uint64_t max_frames,
|
||||
continue;
|
||||
}
|
||||
oamd::OamdUpdate update;
|
||||
if (!oamd::parse_id11(bytes.data(), bytes.size(), &update).ok() ||
|
||||
if (!oamd::parse_id11_state(bytes.data(), bytes.size(), &update).ok() ||
|
||||
!update.has_trim) {
|
||||
continue;
|
||||
}
|
||||
@@ -1429,6 +1429,11 @@ Status run(const joc_task_config& raw, const joc_event_sink* sink, joc_task_resu
|
||||
return finalize_with_error(seeded.code(), "render", seeded.message());
|
||||
}
|
||||
speaker.controls_ready = true;
|
||||
// The metadata path needs the programme trim before its own event
|
||||
// arrives; the probe above read it from the leading frames.
|
||||
speaker.trim = initial_trim;
|
||||
speaker.has_trim = true;
|
||||
speaker.raw_trim_mode = initial_trim.global_trim_mode;
|
||||
}
|
||||
bus.emit(JOC_EV_RENDERER_INITIALIZED, JOC_STAGE_RENDER, JOC_LOG_INFO,
|
||||
fmt("speaker renderer: %s, %u channels", speaker.layout.name,
|
||||
@@ -1589,7 +1594,11 @@ Status run(const joc_task_config& raw, const joc_event_sink* sink, joc_task_resu
|
||||
// count was clamped to 15, and frames whose syncframe carried a second EMDF
|
||||
// container that is not this frame's JOC metadata.
|
||||
std::uint64_t missing_joc_frames = 0;
|
||||
std::uint64_t truncated_joc_frames = 0;
|
||||
std::uint64_t objects_clamped_frames = 0;
|
||||
std::uint64_t bad_parameter_band_frames = 0;
|
||||
std::uint64_t bad_offset_ts_frames = 0;
|
||||
std::uint64_t trailing_data_frames = 0;
|
||||
std::uint64_t extra_container_frames = 0;
|
||||
std::uint64_t chanmap_pair_mismatch_frames = 0;
|
||||
std::uint64_t bed_mask_mismatch_frames = 0;
|
||||
@@ -1660,6 +1669,22 @@ Status run(const joc_task_config& raw, const joc_event_sink* sink, joc_task_resu
|
||||
if (params.objects_clamped != 0u) {
|
||||
++objects_clamped_frames;
|
||||
}
|
||||
switch (params.status) {
|
||||
case JOC_FRAME_STATUS_BAD_PARAMETER_BAND:
|
||||
++bad_parameter_band_frames;
|
||||
break;
|
||||
case JOC_FRAME_STATUS_BAD_OFFSET_TS:
|
||||
++bad_offset_ts_frames;
|
||||
break;
|
||||
case JOC_FRAME_STATUS_TRAILING_DATA:
|
||||
++trailing_data_frames;
|
||||
break;
|
||||
case JOC_FRAME_STATUS_TRUNCATED:
|
||||
++truncated_joc_frames;
|
||||
break;
|
||||
default:
|
||||
break;
|
||||
}
|
||||
} else {
|
||||
++missing_joc_frames;
|
||||
}
|
||||
@@ -1691,8 +1716,9 @@ Status run(const joc_task_config& raw, const joc_event_sink* sink, joc_task_resu
|
||||
}
|
||||
std::string dsp_error;
|
||||
const Status rebuilt = ::joc::joc::rebuild_objects16(
|
||||
rebuilder, params, bed5.data(), lfe.data(), gain, &objects16,
|
||||
&dsp_error);
|
||||
rebuilder, params, bed5.data(), lfe.data(),
|
||||
static_cast<std::uint32_t>(resolved_core_channels), has_joc, gain,
|
||||
&objects16, &dsp_error);
|
||||
if (!rebuilt.ok()) {
|
||||
return Status::fail(rebuilt.code(), "joc", dsp_error);
|
||||
}
|
||||
@@ -1819,8 +1845,17 @@ Status run(const joc_task_config& raw, const joc_event_sink* sink, joc_task_resu
|
||||
stream_done = true;
|
||||
break;
|
||||
}
|
||||
// The speaker, binaural and ADM paths consume the full per-object
|
||||
// block state (multi-block and every declared object); the legacy
|
||||
// stream/ADM single-block view only needs the slot fields.
|
||||
const bool full_metadata_state =
|
||||
speaker_enabled || binaural_enabled || adm_enabled;
|
||||
const Status oamd =
|
||||
::joc::oamd::parse_id11(bytes.data(), bytes.size(), &job.update);
|
||||
full_metadata_state
|
||||
? ::joc::oamd::parse_id11_state(bytes.data(), bytes.size(),
|
||||
&job.update)
|
||||
: ::joc::oamd::parse_id11(bytes.data(), bytes.size(),
|
||||
&job.update);
|
||||
if (!oamd.ok()) {
|
||||
failures.record(produced, StageRank::kJoc, oamd.code(), oamd.stage(),
|
||||
"frame " + std::to_string(produced) + ": " +
|
||||
@@ -2166,11 +2201,33 @@ Status run(const joc_task_config& raw, const joc_event_sink* sink, joc_task_resu
|
||||
joiner.shutdown(false);
|
||||
if (missing_joc_frames != 0u) {
|
||||
bus.warning(std::to_string(missing_joc_frames) +
|
||||
" audio frame(s) carried no JOC payload; their object channels stay silent");
|
||||
" audio frame(s) carried no JOC payload; their core PCM was passed "
|
||||
"through (8 core planes, 1217-sample delay)");
|
||||
}
|
||||
if (objects_clamped_frames != 0u) {
|
||||
bus.warning(std::to_string(objects_clamped_frames) +
|
||||
" audio frame(s) declared more than 15 objects; the count was clamped to 15");
|
||||
" audio frame(s) declared more than 15 objects; the count was clamped "
|
||||
"to 15 and the frame took the invalid-frame path");
|
||||
}
|
||||
if (bad_parameter_band_frames != 0u) {
|
||||
bus.warning(std::to_string(bad_parameter_band_frames) +
|
||||
" audio frame(s) declared an unknown JOC parameter-band count; the "
|
||||
"frame took the invalid-frame path");
|
||||
}
|
||||
if (bad_offset_ts_frames != 0u) {
|
||||
bus.warning(std::to_string(bad_offset_ts_frames) +
|
||||
" audio frame(s) carried an invalid JOC offset_ts value; the frame "
|
||||
"took the invalid-frame path");
|
||||
}
|
||||
if (trailing_data_frames != 0u) {
|
||||
bus.warning(std::to_string(trailing_data_frames) +
|
||||
" audio frame(s) carried JOC data after the byte-alignment padding; "
|
||||
"the frame took the invalid-frame path");
|
||||
}
|
||||
if (truncated_joc_frames != 0u) {
|
||||
bus.warning(std::to_string(truncated_joc_frames) +
|
||||
" audio frame(s) carried a JOC payload that ends inside joc_data; "
|
||||
"the frame took the invalid-frame path and held the previous matrices");
|
||||
}
|
||||
if (extra_container_frames != 0u) {
|
||||
bus.warning(std::to_string(extra_container_frames) +
|
||||
|
||||
@@ -95,37 +95,100 @@ Status ObjectPositionTrack::position_at(std::int64_t sample, double out[3]) {
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
OamdPositionTimeline::OamdPositionTimeline(int object_count) : object_count_(object_count) {
|
||||
if (object_count != kTimelineObjects) {
|
||||
object_count_ = kTimelineObjects;
|
||||
OamdPositionTimeline::OamdPositionTimeline(int object_count)
|
||||
: object_count_(std::clamp(object_count, 1, oamd::kMaxObjects - 1)) {
|
||||
const std::size_t count = static_cast<std::size_t>(object_count_);
|
||||
tracks_.resize(count);
|
||||
previous_targets_.resize(count);
|
||||
has_previous_.assign(count, false);
|
||||
fixed_.assign(count, false);
|
||||
fixed_xyz_.resize(count);
|
||||
}
|
||||
|
||||
void OamdPositionTimeline::ensure_tracks(int count) {
|
||||
const int capped = std::clamp(count, 1, oamd::kMaxObjects - 1);
|
||||
if (capped <= object_count_) {
|
||||
return;
|
||||
}
|
||||
const std::size_t size = static_cast<std::size_t>(capped);
|
||||
object_count_ = capped;
|
||||
tracks_.resize(size);
|
||||
previous_targets_.resize(size);
|
||||
has_previous_.resize(size, false);
|
||||
fixed_.resize(size, false);
|
||||
fixed_xyz_.resize(size);
|
||||
}
|
||||
|
||||
void OamdPositionTimeline::set_fixed_position(int object_index, const double position[3]) {
|
||||
if (object_index < 1 || object_index > object_count_) {
|
||||
return;
|
||||
}
|
||||
const std::size_t index = static_cast<std::size_t>(object_index - 1);
|
||||
fixed_[index] = true;
|
||||
for (int axis = 0; axis < 3; ++axis) {
|
||||
fixed_xyz_[index][static_cast<std::size_t>(axis)] = position[axis];
|
||||
}
|
||||
}
|
||||
|
||||
Status OamdPositionTimeline::submit_update(const oamd::OamdUpdate& update,
|
||||
std::int64_t frame_start_sample,
|
||||
std::int64_t outer_sample_offset,
|
||||
std::int64_t object_delay_samples,
|
||||
std::int64_t processed_sample) {
|
||||
if (frame_start_sample < 0 || outer_sample_offset < 0 || object_delay_samples < 0) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kOamd,
|
||||
"OAMD frame, outer offset and object delay must be non-negative");
|
||||
void OamdPositionTimeline::reset() {
|
||||
state_.reset();
|
||||
for (int index = 0; index < object_count_; ++index) {
|
||||
tracks_[static_cast<std::size_t>(index)] = ObjectPositionTrack();
|
||||
has_previous_[static_cast<std::size_t>(index)] = false;
|
||||
previous_targets_[static_cast<std::size_t>(index)] = {0.0, 0.0, 0.0};
|
||||
}
|
||||
initialized_ = false;
|
||||
payload_count_ = 0;
|
||||
transition_count_ = 0;
|
||||
last_coded_event_sample_ = -1;
|
||||
}
|
||||
state_.apply(update);
|
||||
|
||||
double targets[kTimelineObjects][3] = {};
|
||||
for (int index = 0; index < kTimelineObjects; ++index) {
|
||||
if (fixed_[index]) {
|
||||
std::memcpy(targets[index], fixed_xyz_[index], sizeof(targets[index]));
|
||||
void OamdPositionTimeline::apply_block(const oamd::FrameState& state,
|
||||
std::uint32_t block_index) {
|
||||
// Track i reads OAMD object i + 1: object 0 is the LFE/bed slot the JOC
|
||||
// renderers do not position. Bed and ISF objects carry no position field,
|
||||
// and an inactive block keeps the previous position, exactly like the
|
||||
// reference frame_update() value construction.
|
||||
for (int track = 0; track < object_count_; ++track) {
|
||||
const int object_index = track + 1;
|
||||
if (object_index >= static_cast<int>(state.objects.size())) {
|
||||
break;
|
||||
}
|
||||
const oamd::ObjectElementState& element =
|
||||
state.objects[static_cast<std::size_t>(object_index)];
|
||||
if (block_index >= element.blocks.size()) {
|
||||
continue;
|
||||
}
|
||||
geometry::q_to_adm_xyz(state_.q(index + 1, 0), state_.q(index + 1, 1),
|
||||
state_.q(index + 1, 2), &targets[index][0], &targets[index][1],
|
||||
&targets[index][2]);
|
||||
const oamd::ObjectRenderState& render =
|
||||
element.blocks[static_cast<std::size_t>(block_index)].render;
|
||||
if (render.not_active || element.kind != oamd::ObjectKind::kDynamic) {
|
||||
continue;
|
||||
}
|
||||
state_.apply_render_state(object_index, render);
|
||||
}
|
||||
}
|
||||
|
||||
const std::int64_t coded_event =
|
||||
frame_start_sample + outer_sample_offset +
|
||||
static_cast<std::int64_t>(update.block_offset_samples);
|
||||
void OamdPositionTimeline::targets(std::vector<std::array<double, 3>>* out) const {
|
||||
out->resize(static_cast<std::size_t>(object_count_));
|
||||
for (int index = 0; index < object_count_; ++index) {
|
||||
const std::size_t slot = static_cast<std::size_t>(index);
|
||||
if (fixed_[slot]) {
|
||||
(*out)[slot] = fixed_xyz_[slot];
|
||||
continue;
|
||||
}
|
||||
double x = 0.0;
|
||||
double y = 0.0;
|
||||
double z = 0.0;
|
||||
geometry::q_to_adm_xyz(state_.q(index + 1, 0), state_.q(index + 1, 1),
|
||||
state_.q(index + 1, 2), &x, &y, &z);
|
||||
(*out)[slot] = {x, y, z};
|
||||
}
|
||||
}
|
||||
|
||||
Status OamdPositionTimeline::schedule(const std::vector<std::array<double, 3>>& targets,
|
||||
std::int64_t coded_event, std::int64_t ramp_samples,
|
||||
std::int64_t object_delay_samples,
|
||||
std::int64_t processed_sample) {
|
||||
if (coded_event < last_coded_event_sample_) {
|
||||
return timeline_fail(JOC_ERR_OAMD_UNSUPPORTED_VARIANT,
|
||||
"non_monotonic_binaural_updates: event sample " +
|
||||
@@ -141,44 +204,105 @@ Status OamdPositionTimeline::submit_update(const oamd::OamdUpdate& update,
|
||||
"sample " + std::to_string(processed_sample) +
|
||||
" had been processed, so sample 0 cannot be backfilled");
|
||||
}
|
||||
for (int index = 0; index < kTimelineObjects; ++index) {
|
||||
tracks_[index].set_initial(targets[index]);
|
||||
std::memcpy(previous_targets_[index], targets[index], sizeof(targets[index]));
|
||||
has_previous_[index] = true;
|
||||
for (int index = 0; index < object_count_; ++index) {
|
||||
const std::size_t slot = static_cast<std::size_t>(index);
|
||||
tracks_[slot].set_initial(targets[slot].data());
|
||||
previous_targets_[slot] = targets[slot];
|
||||
has_previous_[slot] = true;
|
||||
}
|
||||
initialized_ = true;
|
||||
++payload_count_;
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
const std::int64_t effective_ramp =
|
||||
std::max<std::int64_t>(0, static_cast<std::int64_t>(update.ramp_duration_samples));
|
||||
const std::int64_t effective_ramp = std::max<std::int64_t>(0, ramp_samples);
|
||||
const std::int64_t transition_start =
|
||||
timing::align_metadata_sample(coded_event + object_delay_samples);
|
||||
for (int index = 0; index < kTimelineObjects; ++index) {
|
||||
const bool changed = !has_previous_[index] ||
|
||||
std::memcmp(previous_targets_[index], targets[index],
|
||||
sizeof(targets[index])) != 0;
|
||||
for (int index = 0; index < object_count_; ++index) {
|
||||
const std::size_t slot = static_cast<std::size_t>(index);
|
||||
const bool changed =
|
||||
!has_previous_[slot] ||
|
||||
std::memcmp(previous_targets_[slot].data(), targets[slot].data(),
|
||||
sizeof(targets[slot])) != 0;
|
||||
if (!changed) {
|
||||
continue;
|
||||
}
|
||||
const Status status = tracks_[index].append(transition_start, effective_ramp,
|
||||
targets[index], index + 1);
|
||||
const Status status =
|
||||
tracks_[slot].append(transition_start, effective_ramp, targets[slot].data(),
|
||||
index + 1);
|
||||
if (!status.ok()) {
|
||||
return status;
|
||||
}
|
||||
std::memcpy(previous_targets_[index], targets[index], sizeof(targets[index]));
|
||||
has_previous_[index] = true;
|
||||
previous_targets_[slot] = targets[slot];
|
||||
has_previous_[slot] = true;
|
||||
++transition_count_;
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
Status OamdPositionTimeline::submit_update(const oamd::OamdUpdate& update,
|
||||
std::int64_t frame_start_sample,
|
||||
std::int64_t outer_sample_offset,
|
||||
std::int64_t object_delay_samples,
|
||||
std::int64_t processed_sample) {
|
||||
if (frame_start_sample < 0 || outer_sample_offset < 0 || object_delay_samples < 0) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kOamd,
|
||||
"OAMD frame, outer offset and object delay must be non-negative");
|
||||
}
|
||||
if (!update.has_frame_state) {
|
||||
state_.apply(update);
|
||||
targets(&targets_);
|
||||
const std::int64_t coded_event =
|
||||
frame_start_sample + outer_sample_offset +
|
||||
static_cast<std::int64_t>(update.block_offset_samples);
|
||||
const Status status =
|
||||
schedule(targets_, coded_event, update.ramp_duration_samples, object_delay_samples,
|
||||
processed_sample);
|
||||
if (status.ok()) {
|
||||
++payload_count_;
|
||||
}
|
||||
return status;
|
||||
}
|
||||
|
||||
const oamd::FrameState& state = update.frame_state;
|
||||
ensure_tracks(state.object_count > 0u ? static_cast<int>(state.object_count) - 1 : 0);
|
||||
const std::uint32_t block_count =
|
||||
state.blocks.empty() ? 1u : static_cast<std::uint32_t>(state.blocks.size());
|
||||
for (std::uint32_t block = 0; block < block_count; ++block) {
|
||||
apply_block(state, block);
|
||||
targets(&targets_);
|
||||
std::int64_t block_offset = 0;
|
||||
std::int64_t ramp_samples = 0;
|
||||
if (block < state.blocks.size()) {
|
||||
block_offset = static_cast<std::int64_t>(state.blocks[block].block_offset_samples);
|
||||
ramp_samples = static_cast<std::int64_t>(state.blocks[block].ramp_duration_samples);
|
||||
}
|
||||
const std::int64_t coded_event =
|
||||
frame_start_sample + outer_sample_offset + block_offset;
|
||||
const Status status = schedule(targets_, coded_event, ramp_samples,
|
||||
object_delay_samples, processed_sample);
|
||||
if (!status.ok()) {
|
||||
return status;
|
||||
}
|
||||
}
|
||||
++payload_count_;
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
Status OamdPositionTimeline::positions_at(std::int64_t sample,
|
||||
double positions[kTimelineObjects][3]) {
|
||||
for (int index = 0; index < kTimelineObjects; ++index) {
|
||||
const Status status = tracks_[index].position_at(sample, positions[index]);
|
||||
Status OamdPositionTimeline::positions_at(std::int64_t sample, double* positions,
|
||||
int count) {
|
||||
if (positions == nullptr || count < 0) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kOamd,
|
||||
"binaural positions need a non-negative count and a buffer");
|
||||
}
|
||||
if (count > object_count_) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kOamd,
|
||||
"binaural positions requested " + std::to_string(count) +
|
||||
" objects but the timeline tracks " +
|
||||
std::to_string(object_count_));
|
||||
}
|
||||
for (int index = 0; index < count; ++index) {
|
||||
const Status status = tracks_[static_cast<std::size_t>(index)].position_at(
|
||||
sample, positions + static_cast<std::size_t>(index) * 3u);
|
||||
if (!status.ok()) {
|
||||
return status;
|
||||
}
|
||||
|
||||
@@ -2,6 +2,7 @@
|
||||
|
||||
#pragma once
|
||||
|
||||
#include <array>
|
||||
#include <cstdint>
|
||||
#include <vector>
|
||||
|
||||
@@ -10,6 +11,11 @@
|
||||
|
||||
namespace joc::timeline {
|
||||
|
||||
// The JOC binaural renderers carry one LFE plus fifteen object sources, but the
|
||||
// timeline is not tied to that product: it keeps one track per declared OAMD
|
||||
// object (bounded only by oamd::kMaxObjects) and the renderer reads as many as
|
||||
// its source layout consumes. A binaural output with a different source count
|
||||
// (for example one fed by a TrueHD decode) can ask for more tracks.
|
||||
inline constexpr int kTimelineObjects = 15;
|
||||
|
||||
struct PositionTransition {
|
||||
@@ -42,56 +48,53 @@ class OamdPositionTimeline {
|
||||
public:
|
||||
explicit OamdPositionTimeline(int object_count = kTimelineObjects);
|
||||
|
||||
// Schedules one OAMD payload. A payload parsed by parse_id11_state() carries
|
||||
// the full frame state: every position block is applied at its own offset and
|
||||
// ramp, and the track list grows to cover every declared object. The legacy
|
||||
// single-block slot view is still accepted for direct callers.
|
||||
Status submit_update(const oamd::OamdUpdate& update, std::int64_t frame_start_sample,
|
||||
std::int64_t outer_sample_offset, std::int64_t object_delay_samples,
|
||||
std::int64_t processed_sample);
|
||||
|
||||
// positions[15][3]; queries must be monotonically increasing.
|
||||
Status positions_at(std::int64_t sample, double positions[kTimelineObjects][3]);
|
||||
// positions[count][3]; queries must be monotonically increasing.
|
||||
Status positions_at(std::int64_t sample, double* positions, int count);
|
||||
|
||||
// bed 声道是固定通道:OAMD 不给位置,方位由声明的 bed 布局给出。
|
||||
void set_fixed_position(int object_index, const double position[3]) {
|
||||
if (object_index < 1 || object_index > kTimelineObjects) {
|
||||
return;
|
||||
}
|
||||
fixed_[object_index - 1] = true;
|
||||
fixed_xyz_[object_index - 1][0] = position[0];
|
||||
fixed_xyz_[object_index - 1][1] = position[1];
|
||||
fixed_xyz_[object_index - 1][2] = position[2];
|
||||
}
|
||||
void set_fixed_position(int object_index, const double position[3]);
|
||||
|
||||
// Drops the OAMD state and every transition but keeps the fixed bed positions:
|
||||
// they describe the content, not the playback position.
|
||||
void reset() {
|
||||
state_.reset();
|
||||
for (int index = 0; index < kTimelineObjects; ++index) {
|
||||
tracks_[index] = ObjectPositionTrack();
|
||||
has_previous_[index] = false;
|
||||
for (int axis = 0; axis < 3; ++axis) {
|
||||
previous_targets_[index][axis] = 0.0;
|
||||
}
|
||||
}
|
||||
initialized_ = false;
|
||||
payload_count_ = 0;
|
||||
transition_count_ = 0;
|
||||
last_coded_event_sample_ = -1;
|
||||
}
|
||||
void reset();
|
||||
|
||||
int object_count() const { return object_count_; }
|
||||
std::uint64_t payload_count() const { return payload_count_; }
|
||||
std::uint64_t transition_count() const { return transition_count_; }
|
||||
bool initialized() const { return initialized_; }
|
||||
|
||||
const ObjectPositionTrack& track(int index) const { return tracks_[index]; }
|
||||
const ObjectPositionTrack& track(int index) const { return tracks_[static_cast<std::size_t>(index)]; }
|
||||
|
||||
private:
|
||||
// Grows the track list so every declared OAMD object has one (the JOC
|
||||
// renderers use the first kTimelineObjects).
|
||||
void ensure_tracks(int count);
|
||||
// OAMD position state after applying one frame-state block.
|
||||
void apply_block(const oamd::FrameState& state, std::uint32_t block_index);
|
||||
// Current ADM cartesian target of every track.
|
||||
void targets(std::vector<std::array<double, 3>>* out) const;
|
||||
// Initialises from the first update or appends the changed transitions.
|
||||
Status schedule(const std::vector<std::array<double, 3>>& targets,
|
||||
std::int64_t coded_event, std::int64_t ramp_samples,
|
||||
std::int64_t object_delay_samples, std::int64_t processed_sample);
|
||||
|
||||
int object_count_;
|
||||
oamd::OamdState state_;
|
||||
ObjectPositionTrack tracks_[kTimelineObjects];
|
||||
std::vector<ObjectPositionTrack> tracks_;
|
||||
bool initialized_ = false;
|
||||
double previous_targets_[kTimelineObjects][3] = {};
|
||||
bool has_previous_[kTimelineObjects] = {};
|
||||
bool fixed_[kTimelineObjects] = {};
|
||||
double fixed_xyz_[kTimelineObjects][3] = {};
|
||||
std::vector<std::array<double, 3>> previous_targets_;
|
||||
std::vector<bool> has_previous_;
|
||||
std::vector<bool> fixed_;
|
||||
std::vector<std::array<double, 3>> fixed_xyz_;
|
||||
std::vector<std::array<double, 3>> targets_;
|
||||
std::uint64_t payload_count_ = 0;
|
||||
std::uint64_t transition_count_ = 0;
|
||||
std::int64_t last_coded_event_sample_ = -1;
|
||||
|
||||
@@ -0,0 +1,161 @@
|
||||
#include "virtual_speaker/virtual_speaker.h"
|
||||
|
||||
#include <algorithm>
|
||||
#include <cmath>
|
||||
#include <cstring>
|
||||
|
||||
#include "joc_core.h"
|
||||
|
||||
namespace joc::virtual_speaker {
|
||||
|
||||
namespace {
|
||||
|
||||
constexpr std::size_t kFrameSamples = JOC_FRAME_SAMPLES;
|
||||
constexpr std::size_t kSourceChannels = JOC_OUTPUT_CHANNELS;
|
||||
|
||||
// WAV order follows the ffmpeg layouts: 5.1 = L R C LFE BL BR,
|
||||
// 7.1 = L R C LFE BL BR SL SR.
|
||||
constexpr Layout kLayouts[] = {
|
||||
{"2.0",
|
||||
2,
|
||||
{{false, -1.0, 1.0, 0.0},
|
||||
{false, 1.0, 1.0, 0.0}}},
|
||||
{"5.1",
|
||||
6,
|
||||
{{false, -1.0, 1.0, 0.0},
|
||||
{false, 1.0, 1.0, 0.0},
|
||||
{false, 0.0, 1.0, 0.0},
|
||||
{true, 0.0, 1.0, 0.0},
|
||||
{false, -1.0, -1.0, 0.0},
|
||||
{false, 1.0, -1.0, 0.0}}},
|
||||
{"7.1",
|
||||
8,
|
||||
{{false, -1.0, 1.0, 0.0},
|
||||
{false, 1.0, 1.0, 0.0},
|
||||
{false, 0.0, 1.0, 0.0},
|
||||
{true, 0.0, 1.0, 0.0},
|
||||
{false, -1.0, -1.0, 0.0},
|
||||
{false, 1.0, -1.0, 0.0},
|
||||
{false, -1.0, 0.0, 0.0},
|
||||
{false, 1.0, 0.0, 0.0}}},
|
||||
};
|
||||
|
||||
std::int16_t q_for_axis(double value) {
|
||||
if (value <= -0.999) {
|
||||
return 0;
|
||||
}
|
||||
if (value >= 0.999) {
|
||||
return 32767;
|
||||
}
|
||||
return 16384;
|
||||
}
|
||||
|
||||
void fill_fixed_positions(const Layout& layout, oamd::OamdUpdate* update) {
|
||||
update->object_count = static_cast<std::uint32_t>(kSourceChannels);
|
||||
update->block_offset_samples = 0;
|
||||
update->ramp_duration_samples = 0;
|
||||
int slot = 1;
|
||||
for (int channel = 0; channel < layout.channels; ++channel) {
|
||||
const ChannelSpec& spec = layout.channel_specs[channel];
|
||||
const int destination = spec.lfe ? 0 : slot++;
|
||||
update->slots[destination].q1 = q_for_axis(spec.x);
|
||||
update->slots[destination].q2 = q_for_axis(-spec.y);
|
||||
update->slots[destination].q3 = q_for_axis(spec.z);
|
||||
}
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
const Layout* find_layout(const std::string& name) {
|
||||
for (const Layout& layout : kLayouts) {
|
||||
if (name == layout.name) {
|
||||
return &layout;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
const Layout* detect_layout(int channels) {
|
||||
for (const Layout& layout : kLayouts) {
|
||||
if (layout.channels == channels) {
|
||||
return &layout;
|
||||
}
|
||||
}
|
||||
return nullptr;
|
||||
}
|
||||
|
||||
Status Input::configure(const std::string& layout_name, int channels, double gain_db) {
|
||||
const Layout* layout =
|
||||
(layout_name.empty() || layout_name == "auto")
|
||||
? detect_layout(channels)
|
||||
: find_layout(layout_name);
|
||||
if (layout == nullptr || layout->channels != channels) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kRender,
|
||||
"PCM layout does not match the channel count");
|
||||
}
|
||||
layout_ = layout;
|
||||
channels_ = channels;
|
||||
gain_ = static_cast<double>(std::pow(10.0, gain_db / 20.0));
|
||||
pending_.clear();
|
||||
read_offset_ = 0;
|
||||
fill_fixed_positions(*layout_, &fixed_update_);
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
void Input::reset() {
|
||||
pending_.clear();
|
||||
read_offset_ = 0;
|
||||
}
|
||||
|
||||
Status Input::append(const float* interleaved, std::size_t frames) {
|
||||
if (interleaved == nullptr || frames == 0u) {
|
||||
return Status::success();
|
||||
}
|
||||
if (layout_ == nullptr || channels_ <= 0) {
|
||||
return Status::fail(JOC_ERR_STATE, stage::kRender,
|
||||
"virtual speaker input is not configured");
|
||||
}
|
||||
pending_.insert(pending_.end(), interleaved,
|
||||
interleaved + frames * static_cast<std::size_t>(channels_));
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
bool Input::frame_ready() const {
|
||||
return layout_ != nullptr &&
|
||||
pending_.size() - read_offset_ >=
|
||||
kFrameSamples * static_cast<std::size_t>(channels_);
|
||||
}
|
||||
|
||||
Status Input::take_frame(std::vector<float>* objects16) {
|
||||
if (objects16 == nullptr) {
|
||||
return Status::fail(JOC_ERR_INVALID_ARGUMENT, stage::kRender,
|
||||
"objects16 output is null");
|
||||
}
|
||||
if (!frame_ready()) {
|
||||
return Status::fail(JOC_ERR_STATE, stage::kRender,
|
||||
"no complete PCM frame is buffered");
|
||||
}
|
||||
objects16->assign(kSourceChannels * kFrameSamples, 0.0f);
|
||||
const float* source = pending_.data() + read_offset_;
|
||||
int slot = 1;
|
||||
for (int channel = 0; channel < layout_->channels; ++channel) {
|
||||
const int destination = layout_->channel_specs[channel].lfe ? 0 : slot++;
|
||||
for (std::size_t sample = 0; sample < kFrameSamples; ++sample) {
|
||||
(*objects16)[static_cast<std::size_t>(destination) * kFrameSamples + sample] =
|
||||
source[sample * static_cast<std::size_t>(channels_) + channel] *
|
||||
static_cast<float>(gain_);
|
||||
}
|
||||
}
|
||||
read_offset_ += kFrameSamples * static_cast<std::size_t>(channels_);
|
||||
if (read_offset_ == pending_.size()) {
|
||||
pending_.clear();
|
||||
read_offset_ = 0;
|
||||
} else if (read_offset_ >= (1u << 20)) {
|
||||
pending_.erase(pending_.begin(),
|
||||
pending_.begin() + static_cast<std::ptrdiff_t>(read_offset_));
|
||||
read_offset_ = 0;
|
||||
}
|
||||
return Status::success();
|
||||
}
|
||||
|
||||
} // namespace joc::virtual_speaker
|
||||
@@ -0,0 +1,57 @@
|
||||
// Virtual speaker input: ordinary multichannel PCM as fixed-direction JOC
|
||||
// objects. It owns only the layout mapping and the PCM-to-objects16 shape;
|
||||
// decoding, speaker rendering and binaural rendering stay in their own modules.
|
||||
#pragma once
|
||||
|
||||
#include <cstddef>
|
||||
#include <cstdint>
|
||||
#include <string>
|
||||
#include <vector>
|
||||
|
||||
#include "foundation/status.h"
|
||||
#include "oamd/oamd_parser.h"
|
||||
|
||||
namespace joc::virtual_speaker {
|
||||
|
||||
struct ChannelSpec {
|
||||
bool lfe;
|
||||
double x;
|
||||
double y;
|
||||
double z;
|
||||
};
|
||||
|
||||
struct Layout {
|
||||
const char* name;
|
||||
int channels;
|
||||
ChannelSpec channel_specs[8];
|
||||
};
|
||||
|
||||
const Layout* find_layout(const std::string& name);
|
||||
const Layout* detect_layout(int channels);
|
||||
|
||||
class Input {
|
||||
public:
|
||||
// gain_db is the module's master attenuation; it is applied to every
|
||||
// mapped channel once, before the objects16 frame leaves the module.
|
||||
Status configure(const std::string& layout_name, int channels, double gain_db = 0.0);
|
||||
void reset();
|
||||
|
||||
const oamd::OamdUpdate& fixed_update() const { return fixed_update_; }
|
||||
int channels() const { return channels_; }
|
||||
const Layout* layout() const { return layout_; }
|
||||
|
||||
// Appends interleaved frames; take_frame() emits whole 1536-frame blocks.
|
||||
Status append(const float* interleaved, std::size_t frames);
|
||||
bool frame_ready() const;
|
||||
Status take_frame(std::vector<float>* objects16);
|
||||
|
||||
private:
|
||||
const Layout* layout_ = nullptr;
|
||||
int channels_ = 0;
|
||||
double gain_ = 1.0;
|
||||
oamd::OamdUpdate fixed_update_{};
|
||||
std::vector<float> pending_;
|
||||
std::size_t read_offset_ = 0;
|
||||
};
|
||||
|
||||
} // namespace joc::virtual_speaker
|
||||
+1053
-11
File diff suppressed because it is too large.
Load diff
@@ -0,0 +1,437 @@
|
||||
// State-only checks of the OAMD parser, ported from tests/oamd_metadata_test.py.
|
||||
// The speaker-renderer targets are covered by the speaker tests; this file keeps
|
||||
// the bitstream expansion, formula tables and programme parsing checks.
|
||||
|
||||
#include <algorithm>
|
||||
#include <cmath>
|
||||
#include <cstdint>
|
||||
#include <cstdio>
|
||||
#include <cstring>
|
||||
#include <vector>
|
||||
|
||||
#include "oamd/oamd_params.h"
|
||||
#include "oamd/oamd_parser.h"
|
||||
|
||||
namespace {
|
||||
|
||||
int failures = 0;
|
||||
|
||||
void check(bool passed, const char* name) {
|
||||
if (!passed) {
|
||||
std::fprintf(stderr, "FAIL: %s\n", name);
|
||||
++failures;
|
||||
}
|
||||
}
|
||||
|
||||
std::vector<std::uint8_t> from_hex(const char* text) {
|
||||
std::vector<std::uint8_t> bytes;
|
||||
const std::size_t length = std::strlen(text);
|
||||
for (std::size_t index = 0; index + 1 < length; index += 2) {
|
||||
const auto high = static_cast<unsigned>(text[index]);
|
||||
const auto low = static_cast<unsigned>(text[index + 1]);
|
||||
const auto digit = [](unsigned value) -> unsigned {
|
||||
if (value >= '0' && value <= '9') {
|
||||
return value - '0';
|
||||
}
|
||||
if (value >= 'a' && value <= 'f') {
|
||||
return value - 'a' + 10;
|
||||
}
|
||||
if (value >= 'A' && value <= 'F') {
|
||||
return value - 'A' + 10;
|
||||
}
|
||||
return 0;
|
||||
};
|
||||
bytes.push_back(static_cast<std::uint8_t>((digit(high) << 4) | digit(low)));
|
||||
}
|
||||
return bytes;
|
||||
}
|
||||
|
||||
joc::oamd::OamdUpdate parse_state(const char* hex) {
|
||||
const std::vector<std::uint8_t> payload = from_hex(hex);
|
||||
joc::oamd::OamdUpdate update;
|
||||
const joc::Status status =
|
||||
joc::oamd::parse_id11_state(payload.data(), payload.size(), &update);
|
||||
check(status.ok(), "parse_id11_state succeeds");
|
||||
if (!status.ok()) {
|
||||
std::fprintf(stderr, " %s: %s\n", hex, status.message().c_str());
|
||||
}
|
||||
return update;
|
||||
}
|
||||
|
||||
const joc::oamd::ObjectRenderState& block0(const joc::oamd::OamdUpdate& update,
|
||||
int object) {
|
||||
return update.frame_state.objects[static_cast<std::size_t>(object)]
|
||||
.blocks[0]
|
||||
.render;
|
||||
}
|
||||
|
||||
// Two object blocks in one object element: block 0 is a full update, block 1 is
|
||||
// an inactive block. Ported from test_multiblock_state_expansion.
|
||||
void test_multiblock_state_expansion() {
|
||||
const joc::oamd::OamdUpdate update = parse_state("0104600800450132831d6a00");
|
||||
check(update.has_frame_state, "frame state present");
|
||||
check(update.frame_state.object_count == 1, "one object");
|
||||
check(update.frame_state.block_count == 2, "two blocks");
|
||||
check(update.frame_state.objects.size() == 1, "one object state");
|
||||
check(update.frame_state.objects[0].blocks.size() == 2, "two object blocks");
|
||||
const joc::oamd::ObjectRenderState& first = block0(update, 0);
|
||||
check(first.gain_q == 23033, "block 0 gain_q");
|
||||
check(first.priority_q == 32767, "block 0 priority_q");
|
||||
check(first.qx == 6342 && first.qy == 21141 && first.qz == -13107, "block 0 position");
|
||||
check(first.zone_idx == 3 && first.enable_elevation, "block 0 zone/elevation");
|
||||
check(first.size_mode == 1, "block 0 size mode");
|
||||
check(first.snap, "block 0 snap");
|
||||
const joc::oamd::ObjectRenderState& second =
|
||||
update.frame_state.objects[0].blocks[1].render;
|
||||
// The second block is active but carries basic_status 0: no gain update.
|
||||
check(!second.not_active, "block 1 active");
|
||||
check(second.gain_q == 0 && second.priority_q == 0, "block 1 gain/priority");
|
||||
check(second.qx == 16384 && second.qy == 16384 && second.qz == 0,
|
||||
"block 1 default position");
|
||||
}
|
||||
|
||||
// Two objects with three blocks: the first object's block 1 is inactive, the
|
||||
// second is a full update. Ported from test_multiblock_gain_inheritance.
|
||||
void test_multiblock_gain_inheritance() {
|
||||
const joc::oamd::OamdUpdate update =
|
||||
parse_state("038474100040850146527514c0407c3e00");
|
||||
check(update.frame_state.objects.size() == 2, "two objects");
|
||||
check(update.frame_state.block_count == 3, "three blocks");
|
||||
const std::int32_t expected_gain[3] = {23033, 0, 2052};
|
||||
for (int block = 0; block < 3; ++block) {
|
||||
check(update.frame_state.objects[0].blocks[static_cast<std::size_t>(block)]
|
||||
.render.gain_q == expected_gain[block],
|
||||
"first object gain inheritance");
|
||||
check(update.frame_state.objects[1].blocks[static_cast<std::size_t>(block)]
|
||||
.render.gain_q == expected_gain[block],
|
||||
"second object gain inheritance");
|
||||
}
|
||||
check(update.frame_state.objects[0].blocks[1].render.not_active,
|
||||
"first object block 1 inactive");
|
||||
check(!update.frame_state.objects[1].blocks[1].render.not_active,
|
||||
"second object block 1 active");
|
||||
for (int block = 0; block < 3; ++block) {
|
||||
check(update.frame_state.objects[0].blocks[static_cast<std::size_t>(block)]
|
||||
.render.qx == 16384,
|
||||
"first object holds centre x");
|
||||
check(update.frame_state.objects[1].blocks[static_cast<std::size_t>(block)]
|
||||
.render.qx == 10570,
|
||||
"second object x");
|
||||
}
|
||||
}
|
||||
|
||||
// gain_idx 0/1/2: unity, mute and the 6-bit dB table.
|
||||
void test_gain_formula() {
|
||||
check(block0(parse_state("010458000501514C0400"), 0).gain_q == 23033, "gain bits 0");
|
||||
check(block0(parse_state("01045800057F514C0400"), 0).gain_q == 14, "gain bits 63");
|
||||
check(block0(parse_state("0104540004D4530100"), 0).gain_q == 0, "gain mute");
|
||||
}
|
||||
|
||||
// size_idx 1/2: isotropic and per-axis Q15 sizes.
|
||||
void test_size_formula() {
|
||||
const joc::oamd::OamdUpdate size16 = parse_state("01045800045453016000");
|
||||
check(block0(size16, 0).size_mode == 1, "size mode 1");
|
||||
check(block0(size16, 0).size_q[0] == 16913, "size value 16");
|
||||
check(block0(size16, 0).size_q[1] == 16913 && block0(size16, 0).size_q[2] == 16913,
|
||||
"isotropic size");
|
||||
const joc::oamd::OamdUpdate size31 = parse_state("01045800045453017E00");
|
||||
check(block0(size31, 0).size_q[0] == 32767, "size value 31");
|
||||
const joc::oamd::OamdUpdate size2 = parse_state("01045C00045453018AAA00");
|
||||
check(block0(size2, 0).size_mode == 2, "size mode 2");
|
||||
check(block0(size2, 0).size_q[0] == 5285 && block0(size2, 0).size_q[1] == 10570 &&
|
||||
block0(size2, 0).size_q[2] == 21141,
|
||||
"per-axis size");
|
||||
}
|
||||
|
||||
// Position quantisation, zone/snap, elevation flag and signed z.
|
||||
void test_position_formula() {
|
||||
const joc::oamd::OamdUpdate base = parse_state("010454000454530100");
|
||||
check(block0(base, 0).qx == 10570 && block0(base, 0).qy == 10570 &&
|
||||
block0(base, 0).qz == 17476,
|
||||
"coarse position");
|
||||
check(block0(base, 0).enable_elevation, "elevation enabled");
|
||||
check(block0(base, 0).zone_idx == 0, "zone 0");
|
||||
|
||||
const joc::oamd::OamdUpdate zone = parse_state("0104540004420A0B10");
|
||||
check(block0(zone, 0).qx == 1057 && block0(zone, 0).qy == 1057 && block0(zone, 0).qz == 0,
|
||||
"zone position");
|
||||
check(block0(zone, 0).zone_idx == 5, "zone index");
|
||||
check(block0(zone, 0).snap, "snap flag");
|
||||
|
||||
const joc::oamd::OamdUpdate height = parse_state("010454000454530000");
|
||||
check(!block0(height, 0).enable_elevation, "elevation disabled");
|
||||
|
||||
const joc::oamd::OamdUpdate negative = parse_state("010454000454510100");
|
||||
check(block0(negative, 0).qz == -17476, "signed elevation");
|
||||
}
|
||||
|
||||
// Extended object element position refinement.
|
||||
void test_refinement() {
|
||||
const joc::oamd::OamdUpdate update = parse_state("010854000454530100A23E00");
|
||||
check(block0(update, 0).qx == 10676, "refined x");
|
||||
check(block0(update, 0).qy == 10676, "refined y");
|
||||
check(block0(update, 0).qz == 17913, "refined z");
|
||||
}
|
||||
|
||||
// Extended object element divergence table.
|
||||
void test_divergence() {
|
||||
check(block0(parse_state("010854000454530100A27400"), 0).divergence_q == 11548,
|
||||
"divergence table value 32");
|
||||
check(block0(parse_state("010854000454530100A277E0"), 0).divergence_q == 16384,
|
||||
"divergence table value 63");
|
||||
}
|
||||
|
||||
// Programme assignment: bed labels, ISF index, masks and distribution flag.
|
||||
void test_program_assignment() {
|
||||
const joc::oamd::OamdUpdate lfe = parse_state("018448000440");
|
||||
check(lfe.frame_state.program.dynamic_object_only, "dynamic object only");
|
||||
check(lfe.frame_state.program.lfe_present, "lfe present");
|
||||
check(lfe.frame_state.program.num_bed_objects == 1, "one bed object");
|
||||
check(lfe.frame_state.program.bed_labels.size() == 1 &&
|
||||
lfe.frame_state.program.bed_labels[0] == joc::joc::BedChannel::kLfe,
|
||||
"lfe bed label");
|
||||
check(lfe.frame_state.program.bed_mask == 0x8u, "lfe bed mask");
|
||||
check(lfe.frame_state.objects[0].kind == joc::oamd::ObjectKind::kBed, "bed object kind");
|
||||
check(block0(lfe, 0).qx == 16384 && block0(lfe, 0).qy == 16384 && block0(lfe, 0).qz == 0,
|
||||
"bed object holds centre");
|
||||
|
||||
const joc::oamd::OamdUpdate isf = parse_state("062011400011084200");
|
||||
check(isf.frame_state.object_count == 4, "isf object count");
|
||||
check(isf.frame_state.program.isf_idx == 0, "isf index");
|
||||
check(isf.frame_state.program.num_isf_objects == 4, "isf object number");
|
||||
for (int index = 0; index < 4; ++index) {
|
||||
check(isf.frame_state.objects[static_cast<std::size_t>(index)].kind ==
|
||||
joc::oamd::ObjectKind::kIsf,
|
||||
"isf object kind");
|
||||
}
|
||||
|
||||
const joc::oamd::OamdUpdate standard = parse_state("021100422600022100");
|
||||
check(standard.frame_state.program.num_bed_objects == 2, "standard bed object number");
|
||||
check(standard.frame_state.program.bed_mask == 0x3u, "standard bed wide mask");
|
||||
check(standard.frame_state.program.bed_labels.size() == 2 &&
|
||||
standard.frame_state.program.bed_labels[0] == joc::joc::BedChannel::kLeft &&
|
||||
standard.frame_state.program.bed_labels[1] == joc::joc::BedChannel::kRight,
|
||||
"standard bed labels");
|
||||
|
||||
const joc::oamd::OamdUpdate nonstandard = parse_state("001000800448000440");
|
||||
check(nonstandard.frame_state.program.num_bed_objects == 1, "non-standard bed number");
|
||||
check(nonstandard.frame_state.program.bed_mask == 0x100u, "non-standard bed mask");
|
||||
check(nonstandard.frame_state.program.bed_labels.size() == 1 &&
|
||||
nonstandard.frame_state.program.bed_labels[0] ==
|
||||
joc::joc::BedChannel::kTopFrontLeft,
|
||||
"non-standard bed label");
|
||||
check(!nonstandard.frame_state.program.b_bed_chan_distribute, "distribution off");
|
||||
|
||||
const joc::oamd::OamdUpdate distributed = parse_state("001800800448000440");
|
||||
check(distributed.frame_state.program.b_bed_chan_distribute, "distribution on");
|
||||
}
|
||||
|
||||
// The parsed bed declaration keeps its expanded channels for the ADM check.
|
||||
void test_bed_assignments() {
|
||||
const std::vector<std::uint8_t> payload = from_hex("021100422600022100");
|
||||
std::vector<joc::oamd::BedAssignment> assignments;
|
||||
check(joc::oamd::parse_id11_bed_assignments(payload.data(), payload.size(),
|
||||
&assignments)
|
||||
.ok(),
|
||||
"bed assignment parse");
|
||||
check(assignments.size() == 1, "one assignment");
|
||||
check(assignments[0].standard && !assignments[0].lfe_only, "standard assignment");
|
||||
check(assignments[0].channels.size() == 2 &&
|
||||
assignments[0].channels[0] == joc::joc::BedChannel::kLeft &&
|
||||
assignments[0].channels[1] == joc::joc::BedChannel::kRight,
|
||||
"assignment channels");
|
||||
}
|
||||
|
||||
// Legacy frame_update() view: 16 slots from block 0, >16 objects and multi-block
|
||||
// frames stay rejected; the full state accepts both.
|
||||
void test_legacy_view() {
|
||||
const std::vector<std::uint8_t> base_payload = from_hex("010454000454530100");
|
||||
joc::oamd::OamdUpdate legacy;
|
||||
check(joc::oamd::parse_id11(base_payload.data(), base_payload.size(), &legacy).ok(),
|
||||
"legacy parse");
|
||||
check(legacy.has_frame_state, "legacy carries frame state");
|
||||
check(legacy.frame_state.block_count == 1, "legacy block count");
|
||||
check(legacy.frame_state.blocks[0].block_offset_samples == 0 &&
|
||||
legacy.frame_state.blocks[0].ramp_duration_samples == 0,
|
||||
"legacy block timing");
|
||||
check(legacy.slots[0].q1 == 10570 && legacy.slots[0].q2 == 10570 &&
|
||||
legacy.slots[0].q3 == 17476,
|
||||
"legacy slot position");
|
||||
check(legacy.slots[1].q1 == -1 && legacy.slots[1].q2 == -1 && legacy.slots[1].q3 == -1,
|
||||
"legacy unused slot hold");
|
||||
|
||||
// The legacy slot view never publishes a negative elevation.
|
||||
const std::vector<std::uint8_t> negative = from_hex("010454000454510100");
|
||||
check(joc::oamd::parse_id11(negative.data(), negative.size(), &legacy).ok(),
|
||||
"negative legacy parse");
|
||||
check(legacy.slots[0].q3 == 0, "legacy negative z clamps to side");
|
||||
|
||||
const std::vector<std::uint8_t> elevation_off = from_hex("010454000454530000");
|
||||
check(joc::oamd::parse_id11(elevation_off.data(), elevation_off.size(), &legacy).ok(),
|
||||
"elevation-off legacy parse");
|
||||
check(legacy.slots[0].q3 == 0, "legacy elevation flag zeroes z");
|
||||
|
||||
// Bed objects have no position field: the slot keeps the previous position.
|
||||
const std::vector<std::uint8_t> bed = from_hex("018448000440");
|
||||
check(joc::oamd::parse_id11(bed.data(), bed.size(), &legacy).ok(), "bed legacy parse");
|
||||
check(legacy.slots[0].q1 == -1 && legacy.slots[0].q2 == -1 && legacy.slots[0].q3 == -1,
|
||||
"legacy bed slot holds");
|
||||
|
||||
const std::vector<std::uint8_t> multiblock = from_hex("0104600800450132831d6a00");
|
||||
const joc::Status legacy_status =
|
||||
joc::oamd::parse_id11(multiblock.data(), multiblock.size(), &legacy);
|
||||
check(!legacy_status.ok() &&
|
||||
legacy_status.code() == JOC_ERR_OAMD_UNSUPPORTED_VARIANT,
|
||||
"legacy rejects multiple blocks");
|
||||
|
||||
const char* many_objects_hex =
|
||||
"21044E600022A29808054530100A8A60201514C0402A29808054530100A8A60201514C040"
|
||||
"2A29808054530100A8A60201514C0402A29808054530100A8A60201514C0402A2980800";
|
||||
const std::vector<std::uint8_t> many_objects = from_hex(many_objects_hex);
|
||||
const joc::Status many_legacy =
|
||||
joc::oamd::parse_id11(many_objects.data(), many_objects.size(), &legacy);
|
||||
check(!many_legacy.ok() && many_legacy.code() == JOC_ERR_OAMD_UNSUPPORTED_VARIANT,
|
||||
"legacy rejects more than 16 objects");
|
||||
joc::oamd::OamdUpdate many;
|
||||
check(joc::oamd::parse_id11_state(many_objects.data(), many_objects.size(), &many).ok(),
|
||||
"full state accepts more than 16 objects");
|
||||
check(many.frame_state.object_count == 17, "full state object count");
|
||||
check(many.frame_state.objects.size() == 17, "full state expands every object");
|
||||
}
|
||||
|
||||
// Optional position fields must be consumed exactly: distance, screen reference
|
||||
// and the priority mask.
|
||||
void test_optional_fields() {
|
||||
const joc::oamd::OamdUpdate distance = parse_state("0304680004545313880D45301000");
|
||||
check(distance.frame_state.object_count == 2, "distance payload object count");
|
||||
check(block0(distance, 0).qx == 10570, "distance payload keeps alignment");
|
||||
check(block0(distance, 1).gain_q == 0, "distance payload second object");
|
||||
|
||||
const joc::oamd::OamdUpdate infinity = parse_state("010454000454531880");
|
||||
check(block0(infinity, 0).qz == 17476, "infinite distance keeps alignment");
|
||||
|
||||
const joc::oamd::OamdUpdate screen = parse_state("01045800045453013600");
|
||||
check(block0(screen, 0).qx == 10570 && block0(screen, 0).snap == false,
|
||||
"screen reference keeps alignment");
|
||||
|
||||
const joc::oamd::OamdUpdate priority = parse_state("03046800043EA29808000A298080");
|
||||
check(block0(priority, 0).priority_q == 31744, "priority bits");
|
||||
check(block0(priority, 1).priority_q == 0, "priority zero");
|
||||
|
||||
const joc::oamd::OamdUpdate inactive = parse_state("010448000600");
|
||||
check(block0(inactive, 0).not_active, "inactive object");
|
||||
check(block0(inactive, 0).gain_q == 0 && block0(inactive, 0).priority_q == 0,
|
||||
"inactive gain/priority");
|
||||
check(block0(inactive, 0).qx == 16384 && block0(inactive, 0).qy == 16384 &&
|
||||
block0(inactive, 0).qz == 0,
|
||||
"inactive default position");
|
||||
}
|
||||
|
||||
// The compiled-in ISF tables must be complete: six configurations, twelve
|
||||
// layout descriptors, four layout groups, 73 paired column flags, and exact
|
||||
// matrix coefficients.
|
||||
void test_metadata_params() {
|
||||
const joc::oamd::MetadataParams* params = joc::oamd::metadata_params();
|
||||
check(params != nullptr, "metadata params");
|
||||
if (params == nullptr) {
|
||||
std::fprintf(stderr, " %s\n", joc::oamd::metadata_params_error());
|
||||
return;
|
||||
}
|
||||
check(params->configs.size() == 6, "six ISF configurations");
|
||||
check(params->layouts.size() == 12, "twelve ISF layout descriptors");
|
||||
check(!params->layout_groups.floor_a.empty() && !params->layout_groups.floor_b.empty() &&
|
||||
!params->layout_groups.height.empty() &&
|
||||
!params->layout_groups.additional.empty(),
|
||||
"four layout groups");
|
||||
check(!params->paired_columns.empty(), "paired column flags");
|
||||
for (const joc::oamd::IsfConfig& config : params->configs) {
|
||||
check(config.counts[4] == static_cast<int>(config.primary_row_map.size()),
|
||||
"isf row count");
|
||||
// Every configuration stores its basis with a 73-column row stride.
|
||||
check(config.matrix.size() ==
|
||||
static_cast<std::size_t>(config.counts[4]) * 73u,
|
||||
"isf matrix size");
|
||||
}
|
||||
const joc::oamd::IsfLayout* layout20 = params->layout("20");
|
||||
const joc::oamd::IsfLayout* layout916 = params->layout("916");
|
||||
check(layout20 != nullptr && layout20->output_count == 2 &&
|
||||
layout20->compensation_value == -4.5,
|
||||
"2.0 layout descriptor");
|
||||
check(layout916 != nullptr && layout916->output_count == 16 &&
|
||||
layout916->compensation_value == 0.0,
|
||||
"9.1.6 layout descriptor");
|
||||
check(params->layout("60") != nullptr && params->layout("60")->output_count == 6 &&
|
||||
params->layout("60")->compensation_value == -1.5,
|
||||
"6.0 layout descriptor");
|
||||
check(!params->layout("unknown"), "unknown layout descriptor");
|
||||
|
||||
// Exact table contents: the config counts and the first coefficients pin
|
||||
// the reference data, and the row/column structure follows from them.
|
||||
const joc::oamd::IsfConfig& first = params->configs[0];
|
||||
check(first.counts[0] == 3 && first.counts[1] == 1 && first.counts[2] == 0 &&
|
||||
first.counts[3] == 0 && first.counts[4] == 4,
|
||||
"first config counts");
|
||||
check(first.matrix.size() == 4u * 73u && first.matrix[0] == 0.806554496f &&
|
||||
first.matrix[1] == 0.55179733f && first.matrix[15] == -0.280463845f,
|
||||
"first config coefficients");
|
||||
check(first.primary_row_map.size() == 4 && first.primary_row_map[3] == 3 &&
|
||||
first.secondary_row_map.size() == 4 && first.secondary_row_map[3] == 3,
|
||||
"first config row maps");
|
||||
const joc::oamd::IsfConfig& last = params->configs[5];
|
||||
check(last.counts[0] == 15 && last.counts[1] == 9 && last.counts[2] == 5 &&
|
||||
last.counts[3] == 1 && last.counts[4] == 30 &&
|
||||
last.matrix.size() == 30u * 73u,
|
||||
"last config shape");
|
||||
|
||||
check(params->layout("20")->output_ids.size() == 37 &&
|
||||
params->layout("20")->output_ids[0] == 0 &&
|
||||
params->layout("20")->output_ids[1] == 1 &&
|
||||
params->layout("20")->output_ids[2] == 0,
|
||||
"layout output ids");
|
||||
check(params->layout_groups.floor_a.size() == 6 &&
|
||||
params->layout_groups.floor_b.size() == 7 &&
|
||||
params->layout_groups.height.size() == 6 &&
|
||||
params->layout_groups.additional.size() == 5,
|
||||
"layout group sizes");
|
||||
check(params->layout_groups.floor_a[1].selector == 1 &&
|
||||
params->layout_groups.floor_a[1].begin == 1 &&
|
||||
params->layout_groups.floor_a[1].end == 3,
|
||||
"layout group record");
|
||||
const std::size_t paired = static_cast<std::size_t>(
|
||||
std::count(params->paired_columns.begin(), params->paired_columns.end(), true));
|
||||
check(params->paired_columns.size() == 73 && paired == 67, "paired column flags");
|
||||
}
|
||||
|
||||
// q_of is the reference's Q15 quantiser.
|
||||
void test_q_of() {
|
||||
check(joc::oamd::q_of(0, 62) == 0, "q_of 0");
|
||||
check(joc::oamd::q_of(62, 62) == 32767, "q_of max");
|
||||
check(joc::oamd::q_of(31, 62) == 16384, "q_of centre");
|
||||
check(joc::oamd::q_of(15, 15) == 32767, "q_of z max");
|
||||
check(joc::oamd::q_of(20, 62) == 10570, "q_of 20/62");
|
||||
check(joc::oamd::q_of(8, 15) == 17476, "q_of 8/15");
|
||||
}
|
||||
|
||||
} // namespace
|
||||
|
||||
int main() {
|
||||
test_multiblock_state_expansion();
|
||||
test_multiblock_gain_inheritance();
|
||||
test_gain_formula();
|
||||
test_size_formula();
|
||||
test_position_formula();
|
||||
test_refinement();
|
||||
test_divergence();
|
||||
test_program_assignment();
|
||||
test_bed_assignments();
|
||||
test_legacy_view();
|
||||
test_optional_fields();
|
||||
test_metadata_params();
|
||||
test_q_of();
|
||||
if (failures != 0) {
|
||||
std::fprintf(stderr, "%d OAMD metadata check(s) failed\n", failures);
|
||||
return 1;
|
||||
}
|
||||
std::printf("OAMD metadata checks passed\n");
|
||||
return 0;
|
||||
}
|
||||
Reference in new issue
Block a user