Files
JustOneCacophony/docs/binaural.en.md
T
TheM14 29f4f83c29
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
Render SOFA through a point-source reference with an optional project room
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

16 KiB
Raw Blame History

Binaural rendering

中文 · Back to README

JustOneCacophony's binaural backend supports three HRTF sources: SimpleFreeFieldHRIR SOFA, the Rosella .personalized_headphone model exported by Dolby's official personalization scan (its JSON parsing is implemented by this project and invokes no Dolby software), and .jochrtf (a JOC compiled HRTF cache).

SOFA keeps its full HRIRs and renders directly through a configured reference environment; it is no longer compiled into a directional field:

SOFA full HRIR [M,2,N]
  -> CanonicalHrtf (48 kHz / one radius shell)
  -> delay-aligned complex spectra (200-1500 Hz phase fit, Data.Delay included)
  -> spherical-triangle barycentric interpolation (lf_logmag_phase or lf_mca_zero)
  -> rigid-sphere radial correction (reference-environment distance)
  -> per-object partitioned convolution (block 128, FFT = max(2048, >=2*support))
  -> reference-environment recursive room model
  -> float64 stereo (one common latency removed)

.jochrtf stays an explicit input (--compiled-hrtf-cache) and keeps the real-SH field renderer; passing --hrtf-cache-policy / --hrtf-cache-dir selects the legacy "SOFA -> .jochrtf" compile path.

Point-source reference scale

Near/Mid/Far use reference distances of 1.00000465 / 2.19327927 / 6.40177584 m; --source-reference-distance-m overrides the selected mode. For an ADM right/front/up position p = (x, y, z):

q_SOFA    = (y, -x, z) / ||p||
d_nominal = reference_distance_m * ||p||
H_render  = H_selected_shell(q_SOFA)

The reference distance only gives object coordinates a nominal metric scale. It does not turn a 1.2 m measurement radius into a 2.193 m ground truth and does not rewrite the SOFA metadata. The default direct mode adds no 1/r attenuation, no head-sphere distance ratio, no synthetic centre response, no room send and no target-distance propagation delay; programme gain and the SOFA response scale are preserved.

A coincident source has no defined direction, so the object keeps its last valid direction, or front when there is no history. Finite positions outside the unit box keep their metadata distance factor and are not clamped.

Optional project room

--sofa-room is off by default and reuses the project's own room algorithms: six first-order image reflections plus a shared FDN tail with a four-channel orthogonal Hadamard feedback matrix, two all-pass diffusers and low-pass damping. It is a listening preset of this project, not a measured BRIR or an external standard room, and it does not use the Rosella room tables.

When enabled, Near/Mid default to the compact room and Far to the legacy room; --sofa-room-preset compact|legacy selects explicitly. The compact geometry follows Zahorik 2009, Table 1 model 9, keeping the original image/FDN algorithm while scaling late delay, diffusion time and RT60 by the V/S ratio; the parameters and their source are recorded in the asset metadata. Only the geometry is borrowed, not that paper's BRIR algorithm.

The full-HRIR route queries an HRIR for each of the six arrival directions and applies the fractional delay as a phase ramp on the full frequency grid; the extra flight time of a reflection is (L_path - L_direct)/c, preserving the original Data.Delay. Direct sound, reflections and the late send use the project's original gain formulas.

The late FDN is rendered at 48 kHz with a unit impulse into a two-channel float64 FIR cut at 2*RT60: 37,383 taps (RT60 about 0.3894 s) for the compact room and 81,600 taps (RT60 0.85 s) for the legacy one. The FIR is produced at startup by the four-line Hadamard feedback loop, the two all-pass diffusers and the low-pass damping in src/binaural/room_tables.cpp; no audio asset ships with the binary. At runtime the same LTI system is applied by PCM-domain FFT convolution in 1536-sample batches, the early reflections still use the selected SOFA, and the LFE channel never enters the room. These states are neither generated nor kept when the room is off, and batch padding is trimmed on flush.

Inputs

A SOFA input renders directly from its full HRIRs and only needs the SOFA file:

joc_cli input.m4a --binaural `
    --sofa-hrtf HRTF\subject.sofa `
    --binaural-mode mid --sofa-interpolation lf_logmag_phase

joc_cli input.m4a --binaural --sofa-hrtf HRTF\subject.sofa --sofa-room
joc_cli input.m4a --binaural --sofa-hrtf HRTF\subject.sofa --sofa-room --sofa-room-preset legacy

A .jochrtf input (and .personalized_headphone) also needs the shared filterbank tables. Both are read-only inputs; parsing and coordinate validation happen while the model is loaded, and nothing is converted again during rendering.

joc_cli input.m4a --binaural `
    --compiled-hrtf-cache path\to\subject.jochrtf `
    --kernels data\rosella_kernels.npz `
    --binaural-mode mid --binaural-tail-seconds 5.0

A multi-radius SOFA requires --hrtf-radius-m; a single shell is selected automatically. --sofa-interpolation picks the local interpolation of the full HRIR: lf_logmag_phase (default, log magnitude plus phase) or lf_mca_zero (auditory-band magnitude correction).

The library exposes the same fields: joc_task_config for a file task, joc_stream_config for streaming, where hrtf_sofa_path (or hrtf_path), sofa_interpolation, hrtf_radius_m, source_reference_distance_m, sofa_room, sofa_room_preset, sofa_room_asset_dir, hrtf_compile, kernels_path, binaural_mode and binaural_tail_seconds configure the binaural path.

The .jochrtf file is an input, not a product of this library: compiling it from SOFA data or measurements belongs to the toolchain and is decoupled from this repository. Loading validates the member set, dtypes and shapes, C-contiguity, CRC-32 and a payload hash recomputed over the members (see .jochrtf below).

binaural_mode is near, mid or far and selects one of the backend's three preset parameter sets; binaural_tail_seconds sets the room-tail length drained on flush (default 5.0 s).

Binaural render mode

--binaural-mode off|near|mid|far (default mid) is a human-specified rendering hint, not original binaural metadata extracted or recovered from the input E-AC-3 JOC bitstream:

  • Direct binaural rendering (--binaural): near/mid/far apply, default mid; off is an error;
  • ADM BWF: the low 3 binaural-render-mode bits of the last 15 JOC object entries in DBMD segment 10 carry off=0/near=1/far=2/mid=3, leaving the first 10 bed entries unchanged; the default is mid, and off explicitly disables the binaural metadata hint.

Canonical SOFA contract

The strict importer currently accepts:

  • Conventions=SOFA;
  • SOFAConventions=SimpleFreeFieldHRIR, version 0.4, 1.0, or 1.1;
  • DataType=FIR and Data.IR[M,2,N];
  • one positive finite Data.SamplingRate in hertz/Hz;
  • spherical or Cartesian SourcePosition;
  • singleton or per-measurement ListenerPosition/View/Up;
  • two receivers whose listener-local lateral geometry uniquely identifies L/R;
  • one zero-offset emitter;
  • causal Data.Delay[I,2] or [M,2];
  • an explicitly free-field/anechoic RoomType.

Receiver order comes from geometry, never from the receiver array index. SOFA listener coordinates are +X front, +Y left, +Z up; ADM coordinates are +X right, +Y front, +Z up:

\bigl(x_{\mathrm{SOFA}},\ y_{\mathrm{SOFA}},\ z_{\mathrm{SOFA}}\bigr) = \bigl(y_{\mathrm{ADM}},\ -x_{\mathrm{ADM}},\ z_{\mathrm{ADM}}\bigr)

CanonicalHrtf keeps Data.IR and Data.Delay separate. Only a time-domain baseline calls materialized_measurement() to apply delay once; the runtime SH path never materializes and then restores the delay. Non-48-kHz HRIRs are normalized with float64 scipy.signal.resample_poly, and delay samples scale by the same ratio.

GeneralFIR, BRIR, TF, multiple emitters, ambiguous receivers, and non-free-field data require convention-specific adapters. They cannot enter the core importer through a reshape.

Exactly-once delay and phase

The compiler recognizes three mutually exclusive representations:

  1. Nonzero Data.Delay is external to Data.IR; the FIR is not de-rotated and runtime applies the delay once.
  2. With Data.Delay=0 and an ordinary positive-onset HRIR, each ear's main peak supplies arrival time. Compilation separates it and runtime restores it once. The current threshold is a peak index greater than two samples.
  3. With Data.Delay=0 and both FIRs at a shared sample-zero origin, no external delay is invented. The authored complex phase stays in the fifth-order field.

No path may add a second ear delay or phase-group delay.

The direct SOFA path does not use this compile-time policy: it keeps the full IRs and estimates the arrival time with a 200-1500 Hz phase fit, so Data.Delay and the fitted delay are each applied once (the delay is removed first and restored afterwards).

Public filterbank and directional field

The SOFA and .jochrtf runtime is fixed at:

  • 48 kHz;
  • a 64-sample QMF hop;
  • 64-QMF / 77 hybrid bands;
  • 961 samples of analysis/synthesis latency;
  • fifth order, 36 terms, ACN/N3D real spherical harmonics;
  • float64 PCM, delay, SH, and room state; complex128 band transfers and spectra.

The .personalized_headphone path selects its bank rather than fixing it. The default is the shared public bank described here (961 samples of latency); bank 2 is the captured Rosella bank (448 samples). Both keep their own coefficients, history and high-band delay, and the runtime skips the selected bank's latency and sizes the flush with it. See Rosella filterbank for the parameters, the CLI switch and the validation results. The SOFA and .jochrtf banks described here are unchanged and never use the selector.

Real and imaginary unit gains for every hybrid band pass through the same analysis/synthesis chain to form a 154-real-parameter impulse dictionary. The 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 hybrid analysis kernels are defined in 3GPP TS 26.405 / ETSI TS 126 405, Section 5.2.2 (Table 1 $Q=8$/ Q=4 coefficients, delay 6):

G_q^p[n] = g^p[n]\cdot\exp\Bigl(j\,\frac{2\pi}{Q^p}\bigl(q+\tfrac12\bigr)(n-6)\Bigr),\qquad n=0,\dots,12

The QMF analysis table is the MPEG-4 AAC/SBR 64 complex QMF bank of ISO/IEC 14496-3/AMD1:2003, subclause 4.B.18.2, stored as the polyphase reordering of the public 640-tap prototype c_0,\dots,c_{639}:

A_{r,t} = \frac{(-1)^t}{128}\,c_{63-r+64t},\qquad r=0,\dots,63,\ t=0,\dots,9

The QMF synthesis table is the causal left inverse of the analysis polyphase matrix \mathbf{A}, i.e. the solution of \mathbf{A}\,\mathbf{W}=\mathbf{P} (\mathbf{P} is the 577-sample delay permutation; total latency 961 = 577 + 6\times64), stored as a rank-4 factorization:

W_{b,l} = \sum_{r=1}^{4} t_{b,l,r}\,\mathbf{b}_{b,r}^{\top}

The hybrid synthesis table is the 77→64 recombination: identity for the high bands, Y_{3+b}=X_{16+b}, and for the low bands( C_p is the 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 and THIRD_PARTY_NOTICES.md for the sources and the rights boundary.

.jochrtf

A .jochrtf file is a pickle-free compressed NumPy archive with an exact member set:

key dtype / shape
metadata_json NumPy Unicode scalar containing JSON text (dtype.kind == "U")
band_center_frequencies_hz little-endian float64[77]
coefficients little-endian complex128[36,2,77]
delay_coefficients little-endian float64[36,2]
delay_bounds little-endian float64[2,2]

Metadata uses the JOC-HRTF-CACHE magic and records the schema, compiler and phase-policy versions, ACN/N3D convention, filterbank hashes, SOFA content SHA-256, sample rate, radius, order, both ridge values, payload hash, and fit report. Every setting that changes compilation participates in the cache key. Metadata never persists an absolute local source_path; it may keep a display name only.

Before constructing a field, the loader uses allow_pickle=False and validates ZIP members and expanded sizes, shapes, dtypes, byte order, contiguous layout, finite values, delay bounds, band centers, payload hash, and cache key. The writer uses a same-directory temporary file, fsync, a process-held OS file lock, and atomic os.replace. Its hidden .lock sidecar may remain and does not mean that a writer still owns the lock. Outdated, damaged, or mismatched caches cannot hit. SOFA input rebuilds an invalid cache; an explicitly selected cache reports the error.

Deleting a disk cache must not change the field or render produced from the same SOFA and compiler configuration.

A .jochrtf file contains directional-field coefficients and delay data transformed from the source HRIRs. Its reproducibility therefore does not make it licence-free. Creating a cache does not enlarge the rights granted by the source SOFA/HRTF dataset: use, copying, and redistribution remain subject to that dataset's terms. If those terms are unclear, keep .jochrtf as a private local cache and do not ship it with the program or another build artifact. source_sha256 is only a content-integrity identifier, not proof of provenance or permission.

JOC objects and room behavior

The production adapter retains the existing JOC schedule:

  • [1536,16] input per frame;
  • channel 0 is special LFE and channels 1..15 are JOC objects;
  • ID11/OAMD positions use a sample-timed timeline;
  • source parameters update every 512 samples;
  • every object owns independent direct/early history while one late FDN is shared;
  • finish() drains early/late tails; output gain is explicit, with no implicit limiter or programme loudness normalization.

Near/Mid/Far, equal-power direct level, six first-order shoebox image sources, late sends, the unitary FDN, the 120–180 Hz cosine-squared LFE low-pass, and room calibration are JOC project-defined behavior, not constants published by SOFA or Dolby.

The binaural renderer is implemented inside this library: the filterbank, the SH directional-field evaluation, the per-object early/direct histories and the shared FDN all run in joc_core (the ejoc_sofa_binaural_* kernel), consuming a compiled directional field and 512-sample metadata updates.

Technical references and rights boundary

Public availability of a specification, source file, or patent document does not by itself authorize copying its contents, redistribution of derivatives, or practice of patent claims. These technical references grant no patent licence and make no non-infringement representation. Anyone preparing a release or product integration must assess the applicable data and software licences, patent permissions, and freedom to operate. See THIRD_PARTY_NOTICES.md for the public-standard provenance and rights boundary.