Commit Graph
13 Commits
Author SHA1 Message Date
TheM14 a2760286b0 Fold the rear surrounds for warp modes 2 and 3
ci / ubuntu-latest (push) Failing after 13s
ci / macos-latest (push) Canceled after 0s
ci / windows-latest (push) Canceled after 0s
ci / release (push) Canceled after 0s
The OAMD warp field is two bits wide; modes 2 and 3 now render instead of
failing. For a 5.1, 5.1.2 or 5.1.4 target the speaker renderer pans objects
in the corresponding 7.1, 7.1.2 or 7.1.4 intermediate layout and folds the
rear pair into the target surround pair: mode 2 uses the PLIIx coefficients
(sqrt(3)/2 and 1/2), mode 3 the Standard (Lo/Ro) coefficients (1 and 0).
4.0 gains a rear pair the same way, built from the 7.1 geometry without a
centre or LFE channel. Layouts that already have both surround pairs, or
none, render directly.

The intermediate layout only changes geometry and panning. Trim
configuration selection, both balance layers and the automatic position
compensation stay with the final target layout, so 5.1.2 keeps configuration
4 and its compensation while panning on 7.1.2. Per-object bypass drops
balance and trim but keeps the fold, and LFE still passes through. The
folded object vector enters the existing gain ramp and mix unchanged; the
public API and the ADM metadata do not change.

parameters_from_oamd accepts the whole two-bit warp field and rejects
values above 3; the renderer builds the fold matrices once per layout and
keeps the 4.0 intermediate geometry internally. The bilingual notes describe
the fold and the evaluation order, and the READMEs no longer call 2/3
unsupported. ctest covers PLIIx and Lo/Ro on 5.1, 5.1.2, 5.1.4 and 4.0, the
target-layout calibration, bypass, LFE and the event path; speaker output
over the shared objects16 stream is byte-identical to the reference for all
layouts in both warp modes 2 and 3.
2026-11-03 13:35:43 +08:00
TheM14 58e3403642 Read the OAMD trim element into speaker rendering and ADM
ci / ubuntu-latest (push) Failing after 16s
ci / macos-latest (push) Canceled after 0s
ci / windows-latest (push) Canceled after 0s
ci / release (push) Canceled after 0s
The OAMD parser reads the trim element: warp mode, global trim mode, the
nine per-configuration records (centre, surround and height level, the two
front-back balances) and the per-object bypass flags. OamdUpdate carries
it, so the file task and the streaming step see the same state; a payload
without the element keeps the previous controls, and an absent element
starts at warp 0 with automatic trim.

Speaker rendering reads the controls from OAMD only; neither the CLI nor
the frontends offer an override. Eligible layouts leave the floor Y alone
at warp 0 and use clamp(2Y, 0, 1) at warp 1, the height layer is never
scaled, and warp 2/3 is reported as unsupported where the speaker
consumes it. Floor, height and trim evaluation keep separate coordinates:
floor Y is balance(warp(Y)), while height Y and trim Y are balance(Y). An
explicit configuration contributes a position-dependent scalar for each
object, applied before accumulation instead of to the output channels
after the mix; automatic, disabled and explicit are mutually exclusive, and
a bypassed object keeps its warp but loses balance and trim. Automatic
still uses the layout position compensation, disabled uses unity, and
layouts without trim semantics stay disabled. LFE passes through, and ADM
source positions and binaural rendering are unchanged.

The controls travel in ejoc_speaker_controls, a C struct a future decoder
can fill without knowing the layout; the layout table now carries the
surround and height channel counts and the seven layouts that have trim
semantics. The internal ABI goes to version 3 for the new process
parameter.

The ADM writer mirrors the element into both carriers. AXML gains
DolbyAtmosMasterFile with DolbyMasterContentInfo, DolbyTrimModeSetup (one
record per configuration, zero values for a disabled one) and
DolbyWarpMode. dbmd segment 10 carries the nine 15-byte records (automatic
flag, three float32 levels and two sign-magnitude balances), and segment 9
writes the warp code into the low three bits of at[152] while keeping the
other packed fields. The task scans the first 64 audio frames for the
element, as the reference does, so a duration-limited render still writes
it.

The bilingual mathematical notes describe the configuration index, the
balance map, the three coordinates and the evaluation order; the READMEs
state that the controls come from OAMD and that warp 2/3 is refused. ctest
covers the element parse, configuration selection, warp 0/1 with the Q15
clamp, the height coordinate, explicit trim and balance, per-object
bypass, LFE passthrough, the step's metadata updates and the exact
dbmd/AXML bytes. Speaker output over the same objects16 stream is
byte-identical to the reference for the seven trimmed layouts and the
wider ones, and both ADM carriers of a rendered BWF are byte-identical to
the reference writer.

The product version goes to 1.2.0; the public ABI stays at 5.
2026-11-03 01:34:15 +08:00
TheM14 20f2e77664 Keep only the Rosella and .jochrtf binaural paths
ci / ubuntu-latest (push) Failing after 18s
ci / macos-latest (push) Canceled after 0s
ci / windows-latest (push) Canceled after 0s
ci / release (push) Canceled after 0s
The kernel renders two inputs now: the Rosella .personalized_headphone model and
an already compiled .jochrtf field. The SOFA reader, its compile cache and the
field fit leave the kernel, and the direct SOFA runtime built on them goes with
them: the point-source distance reference, the shoebox image-source model, the
generated FDN tables and the asset archives are all gone. Fitting a SOFA into a
.jochrtf belongs to the offline toolchain.

The compiled-field runtime is the one that predates that work: 64-QMF analysis,
77 hybrid bands, per-object path delays and the project room (image sources plus
the shared unitary FDN) computed in code. It keeps its behaviour and takes the
.jochrtf name throughout: sofa_binaural_renderer.cpp -> jochrtf_renderer.cpp,
binaural_runtime.* -> jochrtf_runtime.*, SofaBinauralRuntime -> JochrtfRuntime,
ejoc_sofa_binaural_* -> ejoc_jochrtf_*.

The room becomes a switch. --jochrtf-room on|off (default on) drives the
enable_early_reflections and enable_late_room flags of the existing room
configuration; off renders the field's direct binaural sound only.
joc_task_config and joc_stream_config drop hrtf_sofa_path, hrtf_cache_dir,
hrtf_cache_policy and hrtf_radius_m, and gain jochrtf_room. The CLI no longer
accepts --sofa-hrtf, --hrtf-cache-policy, --hrtf-cache-dir or --hrtf-radius-m;
without an explicit input the renderer uses HRTF/binaural.personalized_headphone,
and a .jochrtf has to be named through --compiled-hrtf-cache. The SOFA-only unit
tests go with the reader.

Rosella parsing, filterbank, room and timing are untouched. Both .jochrtf modes
reproduce the Python reference sample for sample, and ctest passes.
2026-11-01 19:19:00 +08:00
TheM14 29f4f83c29 Render SOFA through a point-source reference with an optional project room
ci / ubuntu-latest (push) Failing after 17s
ci / macos-latest (push) Canceled after 0s
ci / windows-latest (push) Canceled after 0s
ci / release (push) Canceled after 0s
Replace the Rosella table-A reference environment with a point-source reference:
the ADM position now selects a listener-frame direction and a distance scaled by
the mode's reference distance (near/mid/far 1.00000465/2.19327927/6.40177584 m,
overridable with --source-reference-distance-m). The direct level is an
equal-power coefficient rather than a 1/r attenuation, and the measured SOFA
response stays as measured: no rigid-sphere radial correction and no Voronoi
centre spectrum any more.

--sofa-room couples the direct path to this project's own room. It is generated
in code, not loaded from an asset: room_tables.cpp runs the four-line Hadamard
FDN with two all-pass diffusers and low-pass damping into a float64 FIR, and
shoebox_room.cpp computes the six first-order image sources, which are rendered
through the same HRIR field with fractional delays. The late tail is convolved
in 1536-sample batches. Near/Mid default to the compact room and Far to the
legacy one; --sofa-room-preset selects explicitly, and the room is off by
default.

Filter changes are scheduled at the parameter alignment delay instead of being
applied on the spot, and the LFE band starts at latency + recording origin - 64,
both matching the reference implementation. joc_task_config goes to version 5
and joc_stream_config gains the same fields; the stream API renders a SOFA
directly instead of compiling it.
2026-11-01 17:20:27 +08:00
TheM14 fd842db26f Render SOFA HRIRs directly and place bed channels by their labels
ci / ubuntu-latest (push) Failing after 11s
ci / macos-latest (push) Canceled after 0s
ci / windows-latest (push) Canceled after 0s
ci / release (push) Canceled after 0s
Two changes that belong to the same reproduction model: a bed channel has
one position, defined by its label, and a SOFA set keeps the impulse
responses it was measured with.

CBA bed channels: the position comes from the channel's label in ETSI
TS 103 420 Table B.10, not from the layout the decoder is asked to render
to. The per-layout anchor table displaced bed channels onto the nearest
target speaker, which silently moved content whenever the target layout
changed.

- Drop the anchor table and the geometry lookup; bed_channel_q15 is the
  only source of bed positions, so stream and task no longer need a
  layout to ask for one
- Keep the 17-bit non-standard bed mask in channel order, as Table 13 /
  TS 103 190-2 Table 65 transmit it
- Exclude LFE channels from the object list while keeping them in the
  coverage mask
- Cover the Table B.10 coordinates, the q15 encoding, the 10-bit and
  17-bit masks and the LFE split in tests

SOFA rendering: compiling a SOFA set into a fifth-order directional field
threw away the measured impulse response - the SH projection, the
64-QMF/77-hybrid round trip and the synthetic room replaced it. The full
IRs are now kept and rendered as such, with the reproduction defined by a
configuration that names a personalized model rather than by the renderer
itself.

- Add SphericalMesh: incremental convex hull over the measurement
  directions, barycentric weights inside the containing triangle and
  spherical Voronoi areas for the centre spectrum
- Add FullHrirField: rfft of the selected shell, 200-1500 Hz phase fit for
  the delay, delay-aligned spectra, log-magnitude/phase or MCA
  interpolation, rigid-sphere radial ratio, recording origin and the
  common FFT/2 latency
- Add Uniform/MultiSourcePartitionedConvolver: block 128, numpy FFT
  conventions, zero-fade filter rebuild from the blocks in flight
- Move the Rosella reference arithmetic (profile geometry, direction
  basis, direct gains and room send, table-A room) out of the renderer
  into rosella_reference so both renderers share it
- Add ReferenceEnvironment and SofaReferenceRuntime: JSON configuration
  with model hash and distance-profile mapping, level calibration over
  equal-area directions, the reference room, 1536-sample batching and a
  single latency removal
- Keep .jochrtf an explicit input and the SOFA compile step behind
  --hrtf-cache-policy/--hrtf-cache-dir; add --reference-config and
  --sofa-interpolation, and make --hrtf-radius-m optional for a single
  shell
- Cover the mesh weights and Voronoi areas, the sphere series and the
  partitioned convolver with reference values in tests
- Document both rules in the READMEs and the binaural notes

Rosella renders stay byte-identical (d5877e89...) and so do renders of
the legacy compiled path (6f0a75d2...). Direct SOFA renders match the
Python reference to 1e-4 in level; the residual comes from the
barycentric solve (analytic inverse here, LAPACK there) on near-degenerate
triangles.
2026-11-01 02:54:55 +08:00
TheM14 34a722f56a Add --rosella-filterbank and restore the public bank as the default
ci / ubuntu-latest (push) Failing after 16s
ci / macos-latest (push) Canceled after 0s
ci / windows-latest (push) Canceled after 0s
ci / release (push) Canceled after 0s
The Rosella path picks its analysis and synthesis bank instead of
hard-coding one. Bank 1 is the shared public bank the renderer used
before the captured variant landed: 13-tap hybrid filters, the 6-slot
high-band delay and 961 samples of latency. Bank 2 is the captured
Rosella bank: 11-tap filters, the 3-slot delay and 448 samples. Both
keep their own coefficients, histories and flush behaviour.

- Drive either bank through one RosellaBank interface, and derive the
  startup skip and the flush capacity from the selected bank
- Add joc_rosella_filterbank to joc_task_config and joc_stream_config
  by replacing the reserved words (struct sizes unchanged, config
  version 3 -> 4)
- Reject an out-of-range value and bank 2 without a Rosella model
  instead of silently falling back
- Expose --rosella-filterbank 1|2 in the CLI with bank 1 as default,
  and reject it outside the personalized headphone path
- Cover bank identity, chunking, reset and per-bank latency in tests
- Document both banks, the selector and the CLI help

Renders of --rosella-filterbank 1 (and the default) match the pre-merge
reference byte for byte; --rosella-filterbank 2 matches the PR #3
reference byte for byte.
2026-10-31 01:11:44 +08:00
TheM14 2a79db4868 Default the nine DBMD configuration trims to automatic
ci / ubuntu-latest (push) Failing after 11s
ci / macos-latest (push) Canceled after 0s
ci / windows-latest (push) Canceled after 0s
ci / release (push) Canceled after 0s
2026-10-29 01:43:56 +08:00
TheM14 7cd5c660e4 Apply each frame's own clipgain in the analysis scalar
ci / ubuntu-latest (push) Failing after 12s
ci / macos-latest (push) Canceled after 0s
ci / windows-latest (push) Canceled after 0s
ci / release (push) Canceled after 0s
2026-10-28 19:20:57 +08:00
TheM14 6f6b5db6c2 Add JOC downmix configurations 0..4, cross-layer notes, and the analysis-side clipgain
ci / ubuntu-latest (push) Failing after 15s
ci / macos-latest (push) Has been cancelled
ci / windows-latest (push) Has been cancelled
2026-10-16 19:52:04 +08:00
TheM14 777a8450f3 Resolve ffmpeg through FFMPEG and PATH instead of a file check
ci / ubuntu-latest (push) Failing after 14s
ci / macos-latest (push) Has been cancelled
ci / windows-latest (push) Has been cancelled
2026-10-14 17:35:28 +08:00
TheM14 9b71940722 Fix the result JSON escaping and indent both JSON outputs
ci / ubuntu-latest (push) Failing after 19s
ci / macos-latest (push) Has been cancelled
ci / windows-latest (push) Has been cancelled
2026-10-09 00:11:40 +08:00
TheM14 9b21f8861c Fix the objects16 input layout and bound its render-ahead
ci / ubuntu-latest (push) Failing after 11s
ci / macos-latest (push) Has been cancelled
ci / windows-latest (push) Has been cancelled
2026-10-06 15:03:49 +08:00
TheM14 42d6da6302 Initial pure C++ implementation
ci / ubuntu-latest (push) Failing after 13s
ci / macos-latest (push) Has been cancelled
ci / windows-latest (push) Has been cancelled
2026-09-25 04:17:30 +08:00