Carry the OAMD trim element into the ADM BWF metadata
Parse the trim element: warp_mode, global_trim_mode, the nine
per-configuration trim_balance_presence records, and the per-object
bypass flags. Select the configuration that applies to a layout from
its surround and height speaker counts:
cfg = kappa_f(surrounds) + 3 * kappa_h(heights)
kappa_f: 0 -> 0, 2..3 -> 1, otherwise 2
kappa_h: 0 -> 0, 2 -> 1, otherwise 2
Write the configuration to both carriers:
* AXML gets a DolbyAtmosMasterFile container holding
DolbyMasterContentInfo, DolbyTrimModeSetup (one records entry per
configuration with centerTrim, surroundTrim, heightTrim and the two
balance fields) and DolbyWarpMode.
* Segment 10 of the dbmd chunk gets the nine 15-byte records: bit 0
is the automatic flag, and an explicit record carries three float32
trims followed by two sign-magnitude balance bytes.
Configurations left at default keep the automatic flag and carry no
value elements.
This commit is contained in:
1 parent
07a26b41fe
commit
220e791ce9
6 files changed
+266
-11
No files matched your search
+6
-3
@@ -71,13 +71,14 @@ class StreamingMaster:
|
||||
"""
|
||||
|
||||
def __init__(self, out_path, duration_sec, rate=48000, block_samples=131072,
|
||||
joc_binaural_mode=4):
|
||||
joc_binaural_mode=4, atmos_meta=None):
|
||||
if block_samples < 1536:
|
||||
raise ValueError("block_samples must be at least one E-AC-3 frame")
|
||||
self.out_path = os.fspath(out_path)
|
||||
self.duration_sec = float(duration_sec)
|
||||
self.rate = int(rate)
|
||||
self.joc_binaural_mode = joc_binaural_mode
|
||||
self.atmos_meta = atmos_meta
|
||||
self._sink = adm_atmos.Sink25(self.out_path, 25, self.rate)
|
||||
self._buffer = np.empty((int(block_samples), 25), dtype=np.float32)
|
||||
self._used = 0
|
||||
@@ -113,10 +114,12 @@ class StreamingMaster:
|
||||
from . import adm_serializer
|
||||
except ImportError:
|
||||
import adm_serializer
|
||||
axml = adm_serializer.build_axml(kf_tracks, self.duration_sec)
|
||||
axml = adm_serializer.build_axml(kf_tracks, self.duration_sec,
|
||||
atmos_meta=self.atmos_meta)
|
||||
chna = adm_atmos.build_chna()
|
||||
dbmd = adm_atmos.build_dbmd(
|
||||
25, joc_binaural_mode=self.joc_binaural_mode)
|
||||
25, joc_binaural_mode=self.joc_binaural_mode,
|
||||
atmos_meta=self.atmos_meta)
|
||||
trajectory_blocks = sum(len(track[1]) for track in kf_tracks)
|
||||
self.metadata_info = {
|
||||
"axml_bytes": len(axml),
|
||||
|
||||
+44
-3
@@ -32,6 +32,10 @@ JOC_BINAURAL_MODES = {
|
||||
JOC_BINAURAL_MODE_DEFAULT = "unspecified"
|
||||
|
||||
def q_to_adm_xyz(q1, q2, q3):
|
||||
"""OAMD 的 q1/q2/q3 编码 → ADM 笛卡尔坐标。
|
||||
|
||||
posX/posY 为 62 级量化,posZ 为 15 级;ADM 的 Y 轴与 OAMD 编码方向相反。
|
||||
"""
|
||||
posX = min(1.0, round(q1 * 62 / 32767.0) / 62.0)
|
||||
posY = min(1.0, round(q2 * 62 / 32767.0) / 62.0)
|
||||
posZ = round(q3 * 15 / 32767.0) / 15.0
|
||||
@@ -195,8 +199,22 @@ def _checksum(seg):
|
||||
s += b
|
||||
return (~s + 1) & 0xFF
|
||||
|
||||
def build_dbmd(object_count=25, joc_binaural_mode=4):
|
||||
"""仅覆盖 segment 10 中 JOC object slots 10..24 的 mode 低 3 bit。"""
|
||||
def _balance_byte(value):
|
||||
"""trim 记录里的 balance 字段:bit7 为符号,bit0-6 为幅度。
|
||||
|
||||
bal3d_y 的取值落在 [0, 1],7 bit 幅度即该区间上的定点表示,步长 1/127。
|
||||
"""
|
||||
value = float(value)
|
||||
magnitude = min(int(round(abs(value) * 127.0)), 127)
|
||||
return magnitude | (0x80 if value < 0 else 0)
|
||||
|
||||
|
||||
def build_dbmd(object_count=25, joc_binaural_mode=4, atmos_meta=None):
|
||||
"""segment 10:9 个 trim 配置 + 逐轨 bypass + JOC object slots 10..24 的 mode。
|
||||
|
||||
含显式值的配置写成显式记录(3 个 float32 + 2 个 balance 字节),其余保持
|
||||
自动标志。
|
||||
"""
|
||||
mode = operator.index(joc_binaural_mode)
|
||||
if mode not in JOC_BINAURAL_MODES.values():
|
||||
raise ValueError(f"invalid JOC binaural render mode: {mode}")
|
||||
@@ -230,7 +248,30 @@ def build_dbmd(object_count=25, joc_binaural_mode=4):
|
||||
# default is automatic, so a freshly rendered master opens with automatic
|
||||
# trims in the renderer.
|
||||
for layout in range(9):
|
||||
ob[7 + 15 * layout] |= 1
|
||||
base = 7 + 15 * layout
|
||||
record = None
|
||||
if atmos_meta is not None:
|
||||
configs = atmos_meta.get("configs") or ()
|
||||
if layout < len(configs):
|
||||
record = configs[layout]
|
||||
explicit = isinstance(record, dict) and not record.get("default")
|
||||
if not explicit:
|
||||
ob[base] |= 1 # bit0 = 1:由渲染器推导
|
||||
continue
|
||||
ob[base] &= 0xFE # bit0 = 0:显式
|
||||
if record.get("disabled"):
|
||||
trims = (0.0, 0.0, 0.0)
|
||||
bal3d_y_tb = bal3d_y_lis = 0.0
|
||||
else:
|
||||
trims = (record.get("centre", 0.0), record.get("surround", 0.0),
|
||||
record.get("height", 0.0))
|
||||
bal3d_y_tb = record.get("bal3d_y_tb", 0.0)
|
||||
bal3d_y_lis = record.get("bal3d_y_lis", 0.0)
|
||||
for index, value in enumerate(trims):
|
||||
offset = base + 1 + 4 * index
|
||||
ob[offset:offset + 4] = struct.pack("<f", float(value))
|
||||
ob[base + 13] = _balance_byte(bal3d_y_tb)
|
||||
ob[base + 14] = _balance_byte(bal3d_y_lis)
|
||||
object_modes = 4 + 2 + 1 + 9 * 15 + object_count
|
||||
for i in range(10, min(object_count, 10 + N_OBJ)):
|
||||
ob[object_modes + i] = (ob[object_modes + i] & 0xF8) | mode
|
||||
|
||||
+70
-2
@@ -23,8 +23,74 @@ def ts(seconds):
|
||||
def esc(v):
|
||||
return (str(v).replace("&", "&").replace("<", "<").replace(">", ">"))
|
||||
|
||||
def build_axml(obj_tracks, duration_sec):
|
||||
"""obj_tracks: [(name, [(rtime, x, y, z, dur), ...]) ×15]"""
|
||||
def _fmt_float(value):
|
||||
return f"{float(value):.6f}"
|
||||
|
||||
|
||||
def _trim_records(atmos_meta):
|
||||
"""把 OAMD trim element 展开成 DolbyTrimModeSetup 的 records。
|
||||
|
||||
``configs`` 每项有三种形态:
|
||||
* ``{"default": True}`` —— 交给渲染器推导,不写任何值子元素;
|
||||
* ``{"disabled": True}`` —— 该配置不施加 trim,写全 0;
|
||||
* 含 centre/surround/height/bal3d_y_tb/bal3d_y_lis 的显式配置。
|
||||
缺省字段一律补 0。
|
||||
"""
|
||||
lines = ["<DolbyTrimModeSetup>"]
|
||||
for index, record in enumerate(atmos_meta.get("configs") or ()):
|
||||
if record is None:
|
||||
continue
|
||||
lines.append("<records>")
|
||||
lines.append(f"<type>{index}</type>")
|
||||
if isinstance(record, dict) and not record.get("default"):
|
||||
if record.get("disabled"):
|
||||
values = {"centerTrim": 0.0, "surroundTrim": 0.0,
|
||||
"heightTrim": 0.0, "balanceOverhead": 0.0,
|
||||
"balanceListener": 0.0}
|
||||
else:
|
||||
values = {
|
||||
"centerTrim": record.get("centre", 0.0),
|
||||
"surroundTrim": record.get("surround", 0.0),
|
||||
"heightTrim": record.get("height", 0.0),
|
||||
"balanceOverhead": record.get("bal3d_y_tb", 0.0),
|
||||
"balanceListener": record.get("bal3d_y_lis", 0.0),
|
||||
}
|
||||
for name, value in values.items():
|
||||
lines.append(f"<{name}>{_fmt_float(value)}</{name}>")
|
||||
lines.append("</records>")
|
||||
lines.append("</DolbyTrimModeSetup>")
|
||||
return lines
|
||||
|
||||
|
||||
def _warp_xml_value(atmos_meta):
|
||||
"""OAMD 的 warp_mode → ADM ``DolbyWarpMode`` 取值。
|
||||
|
||||
warp_mode 选择下混/渲染模式,四档依次为 Direct render、Direct render
|
||||
with room balance、Dolby Pro Logic IIx、Standard (Lo/Ro),ADM 沿用同一
|
||||
编号,因此直接照搬码流值。
|
||||
"""
|
||||
return max(0, min(3, int(atmos_meta.get("warp_mode", 0))))
|
||||
|
||||
|
||||
def _atmos_master_block(atmos_meta):
|
||||
"""Dolby 扩展容器:内容信息 + trim 记录 + warp。"""
|
||||
lines = ["<DolbyAtmosMasterFile>"]
|
||||
warp = _warp_xml_value(atmos_meta)
|
||||
lines.append("<DolbyMasterContentInfo>")
|
||||
lines.append(f"<render_warp_mode>{warp}</render_warp_mode>")
|
||||
lines.append("</DolbyMasterContentInfo>")
|
||||
lines.extend(_trim_records(atmos_meta))
|
||||
lines.append(f"<DolbyWarpMode>{warp}</DolbyWarpMode>")
|
||||
lines.append("</DolbyAtmosMasterFile>")
|
||||
return lines
|
||||
|
||||
|
||||
def build_axml(obj_tracks, duration_sec, atmos_meta=None):
|
||||
"""obj_tracks: [(name, [(rtime, x, y, z, dur), ...]) ×15]
|
||||
|
||||
``atmos_meta`` 为 ``oamd_bits.parse_trim`` 的结果;给出时把 trim 与 warp
|
||||
写进 ADM。
|
||||
"""
|
||||
w = []
|
||||
a = w.append
|
||||
a('<?xml version="1.0" encoding="utf-8"?>')
|
||||
@@ -131,6 +197,8 @@ def build_axml(obj_tracks, duration_sec):
|
||||
a(f'<audioPackFormatIDRef>AP_0003{0x1001 + i:04x}</audioPackFormatIDRef>')
|
||||
a(f'<audioTrackFormatIDRef>AT_0003{0x1001 + i:04x}_01</audioTrackFormatIDRef>')
|
||||
a('</audioStreamFormat>')
|
||||
if atmos_meta:
|
||||
w.extend(_atmos_master_block(atmos_meta))
|
||||
a('</audioFormatExtended></format></coreMetadata>')
|
||||
a('</ebuCoreMain>')
|
||||
return ''.join(w).encode('utf-8')
|
||||
+125
-1
@@ -18,6 +18,13 @@ RAMP_DURATION_INDEX = (
|
||||
1024, 1600, 1601, 1602, 1920, 2000, 2002, 2048,
|
||||
)
|
||||
OBJECT_ELEMENT_ID = 1
|
||||
TRIM_ELEMENT_ID = 2
|
||||
# Table 12 of ETSI TS 103 420:trim 电平表,4 bit 索引,单位 dB。
|
||||
# 索引 15 的 -36 dB 是最小值(并非静音)。
|
||||
TRIM_LUT = (6.0, 3.0, 1.5, 0.75, -0.75, -1.5, -3.0, -4.5,
|
||||
-6.0, -7.5, -9.0, -10.5, -12.0, -13.5, -15.0, -36.0)
|
||||
# bal3d_y 的编码为 (code + 1) / 16,幅度落在 [1/16, 1],取不到 0。
|
||||
BALANCE_STEP = 16
|
||||
# 5.6.0.11 Table 11b:ISF 类型 → 对象数。
|
||||
ISF_OBJECT_COUNTS = (4, 8, 10, 14, 15, 30)
|
||||
# 5.6.1.1.4 Table 12:10 bit 标准 bed 掩码。array position 9(RC_L/RC_R)是 channel
|
||||
@@ -371,6 +378,118 @@ def _parse_object_element(reader, element, raw_payload, retain_spatial=False):
|
||||
}
|
||||
|
||||
|
||||
def _parse_trim_element(reader, object_count, raw_payload):
|
||||
"""``trim_element``:全局 trim/warp 配置,每个 payload 出现一次。
|
||||
|
||||
该 element 描述整段节目的渲染校准意图,不随 block 变化,因此解析结果应当
|
||||
在 payload 之间保持,而不是逐帧重算。
|
||||
"""
|
||||
trim = {
|
||||
"warp_mode": reader.read(2),
|
||||
"reserved": reader.read(2),
|
||||
"global_trim_mode": reader.read(2),
|
||||
"configs": [None] * 9,
|
||||
"disable_per_obj": None,
|
||||
}
|
||||
# global_trim_mode == 2 才携带 9 个配置的显式数据;其余取值表示内容
|
||||
# 把 trim 交给渲染器自行推导(auto)。
|
||||
if trim["global_trim_mode"] == 2:
|
||||
for config in range(9):
|
||||
if reader.read(1): # b_default_trim
|
||||
trim["configs"][config] = {"default": True}
|
||||
continue
|
||||
if reader.read(1): # b_disable_trim
|
||||
trim["configs"][config] = {"disabled": True}
|
||||
continue
|
||||
present = reader.read(5)
|
||||
entry = {"present": present}
|
||||
if present & 0x01:
|
||||
entry["centre"] = TRIM_LUT[reader.read(4)]
|
||||
if present & 0x02:
|
||||
entry["surround"] = TRIM_LUT[reader.read(4)]
|
||||
if present & 0x04:
|
||||
entry["height"] = TRIM_LUT[reader.read(4)]
|
||||
if present & 0x08:
|
||||
sign = 1.0 if reader.read(1) else -1.0
|
||||
entry["bal3d_y_tb"] = (
|
||||
(reader.read(4) + 1) / BALANCE_STEP * sign)
|
||||
if present & 0x10:
|
||||
sign = 1.0 if reader.read(1) else -1.0
|
||||
entry["bal3d_y_lis"] = (
|
||||
(reader.read(4) + 1) / BALANCE_STEP * sign)
|
||||
trim["configs"][config] = entry
|
||||
if reader.read(1): # b_disable_trim_per_obj
|
||||
trim["disable_per_obj"] = [bool(reader.read(1))
|
||||
for _ in range(object_count)]
|
||||
return trim
|
||||
|
||||
|
||||
def parse_trim(bits_one):
|
||||
"""独立解析一个 OAMD payload 的 trim element。
|
||||
|
||||
返回 ``None`` 表示该 payload 不含 trim element。异常统一转成
|
||||
``UnsupportedVariantError``,与其它解析入口一致。
|
||||
"""
|
||||
bits, raw_payload = _payload_bits(bits_one)
|
||||
try:
|
||||
version = bits[0] << 1 | bits[1]
|
||||
except IndexError:
|
||||
raise UnsupportedVariantError(
|
||||
"oamd", "header_truncated", "OAMD payload 不足以容纳 header",
|
||||
details={"payload_bits": len(bits),
|
||||
"payload": bytes_descriptor(raw_payload)}) from None
|
||||
reader = _BitReader(bits, 2)
|
||||
try:
|
||||
if version == 3:
|
||||
version += reader.read(3)
|
||||
object_count_bits = reader.read(5)
|
||||
if object_count_bits == 0x1F:
|
||||
object_count_bits += reader.read(7)
|
||||
object_count = object_count_bits + 1
|
||||
program = _parse_program_assignment(reader, raw_payload)
|
||||
except UnsupportedVariantError:
|
||||
raise
|
||||
except ValueError as exc:
|
||||
raise _syntax_error(
|
||||
"header_truncated", "OAMD header 字段越界或不完整",
|
||||
raw_payload, exc, payload_bits=len(bits)) from exc
|
||||
alternate_present = bool(reader.read(1))
|
||||
try:
|
||||
element_count = reader.read(4)
|
||||
except ValueError as exc:
|
||||
raise _syntax_error(
|
||||
"element_count_truncated", "OAMD element 数量字段不完整",
|
||||
raw_payload, exc) from exc
|
||||
for ordinal in range(element_count):
|
||||
header_start = reader.position
|
||||
try:
|
||||
element_id = reader.read(4)
|
||||
size_bytes = _variable_bits_max(reader, 4, 4) + 1
|
||||
except ValueError as exc:
|
||||
raise _syntax_error(
|
||||
"element_header_truncated", "OAMD element header 不完整",
|
||||
raw_payload, exc,
|
||||
element_ordinal=ordinal, header_start_bit=header_start) from exc
|
||||
region_start = reader.position
|
||||
region_end = region_start + size_bytes * 8
|
||||
if region_end > reader.limit:
|
||||
raise UnsupportedVariantError(
|
||||
"oamd", "element_bounds",
|
||||
"OAMD element 声明长度超过 payload 边界",
|
||||
details={"element_ordinal": ordinal, "element_id": element_id,
|
||||
"size_bytes": size_bytes,
|
||||
"payload": bytes_descriptor(raw_payload)})
|
||||
if alternate_present:
|
||||
reader.read(4)
|
||||
reader.read(1)
|
||||
if element_id == TRIM_ELEMENT_ID:
|
||||
trim = _parse_trim_element(reader, object_count, raw_payload)
|
||||
trim["object_count"] = object_count
|
||||
return trim
|
||||
reader.position = region_end
|
||||
return None
|
||||
|
||||
|
||||
def _parse_elements(reader, alternate_present, object_count, program, raw_payload, retain_spatial=False):
|
||||
"""5.5.4 ``oa_element_md()`` 目录;对象 element 按结构解析,其余按声明长度跳过。"""
|
||||
try:
|
||||
@@ -453,8 +572,13 @@ def _parse_elements(reader, alternate_present, object_count, program, raw_payloa
|
||||
element["object_element"] = object_element
|
||||
# 个别编码器声明的 oa_element_size 比实际内容短;以结构解析结果为准。
|
||||
reader.position = max(region_end, reader.position)
|
||||
elif element_id == TRIM_ELEMENT_ID:
|
||||
element["trim_element"] = _parse_trim_element(
|
||||
reader, object_count, raw_payload)
|
||||
element["parsed_end_bit"] = reader.position
|
||||
reader.position = max(region_end, reader.position)
|
||||
else:
|
||||
# trim / extended / 未知 element:按声明长度整体跳过(5.5.4)。
|
||||
# extended / 未知 element:按声明长度整体跳过(5.5.4)。
|
||||
element["parsed_end_bit"] = region_end
|
||||
reader.position = region_end
|
||||
elements.append(element)
|
||||
|
||||
+18
-1
@@ -3,7 +3,7 @@ import math
|
||||
|
||||
from adm_atmos import q_to_adm_xyz
|
||||
from bed_layout import bed_object_positions
|
||||
from oamd_bits import JocFieldState, frame_update
|
||||
from oamd_bits import JocFieldState, frame_update, parse_trim
|
||||
from variant_error import UnsupportedVariantError
|
||||
|
||||
# 元数据更新时刻按渲染器处理块对齐,与 Dolby 的 processing_block_size 及
|
||||
@@ -192,6 +192,23 @@ def _expand_events(events, total_samples, rate, update_block_samples,
|
||||
object_delay_samples, object_index)
|
||||
raise ValueError(f"未知 trajectory_mode: {trajectory_mode}")
|
||||
|
||||
def read_atmos_meta(index, frames=None, max_scan=64):
|
||||
"""从 OAMD payload 读取 trim/warp 配置。
|
||||
|
||||
trim element 描述整段节目的校准意图,跨帧恒定,因此扫到第一帧含 trim 的
|
||||
payload 即可。返回 ``None`` 表示该节目没有 trim element。
|
||||
"""
|
||||
rows = index.rows if frames is None else frames
|
||||
for row in rows[:max_scan]:
|
||||
subs = index.subpayloads(row)
|
||||
if not subs or 11 not in subs:
|
||||
continue
|
||||
meta = parse_trim(subs[11])
|
||||
if meta is not None:
|
||||
return meta
|
||||
return None
|
||||
|
||||
|
||||
def build_adm_tracks(index, frames=None, rate=48000, frame_samples=1536,
|
||||
update_block_samples=METADATA_BLOCK_SAMPLES,
|
||||
object_delay_samples=1473, trajectory_mode="compact"):
|
||||
|
||||
Reference in new issue
Block a user