Matiq/soothe2-re
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
265 Commits 1 Branch 1 Tag
3411e9b42e9c96e61ea678b1fdd5d4463ff1c171
Commit Graph
75 Commits
Author SHA1 Message Date
Matiq 0d262461ea P4: lock bigkernel semantics (exp2(-mask)*blend) + calibrate scale 
Read the 0x26b820 SIMD loop: the exp2 result is multiplied by the blend buffer
(vmulpd at 0x18026bba0 with ymm11=blend) => mask = exp2(-level)*blend, sign
inverted for attenuation. framed_model.cpp corrected to exp2(-scratch)*0.8.
Calibrated level_scale=600 -> t1kq err -0.06 dB (ref -14.98). comb per-tone
errors reduced (max ~7.8 dB); warp/FFT-conv smoothing still approximated.
 2026-08-20 15:25:49 +03:00
Matiq d4a0a68584 P4: transcribe exact FUN_180529fe0 mask chain (structural, not yet calibrated) 
framed_model.cpp now follows the decoded mono-path structure:
  level -> scale(0x540870*0x54088c/0x1a0) -> IIR1 leaky(A1/B1) -> IIR2(A2/B2)
  -> exp2(0.5*(mask-blend)) -> combine/acc -> warp -> IIR3(A3/B3)x2 -> dry/wet
IIR stages are the real live tables (iir_leaky y=A*acc+B*x, B=1-A). Removed the
empirical mask_lut=(1/(1+K*acc))^n. Not numerically calibrated yet: exp2 bigkernel
exact semantics + PRNG prologue (fVar30) + FFT-conv still TBD; t1kq depth -2.9 vs
-14.98 dB ref.
 2026-08-20 15:10:29 +03:00
Matiq f23bbfa1f6 P4: decode FUN_180529fe0 mono-path exactly + extract live mask tables 
Full decomp of the per-band mask-apply loop (scale -> IIR1/2/3 leaky ->
blend -> exp2 bigkernel -> combine/accumulate -> dual warp -> dry/wet -> FFT-conv).
Extracted runtime tables to rt_mask_tables.{hpp,cpp}: IIR A1/B1,A2/B2,A3/B3
(leaky y=A*acc+B*x, B=1-A), warp 0x5406a8, per-band 0x540768, PRNG LUT 0x5408b0.
bigkernel 0x26b820 = vectorized exp2 (log2e/floor/mantissa tables). Fixed the
broken warp line in framed_model.cpp (compiles again).
 2026-08-20 15:00:53 +03:00
Matiq fc48a9fee4 P4: real mask chain in FramedDetector (level-tracker + accumulator + live-calibrated LUT), t1kq err 0.5dB 
Replace empirical PCHIP detector with live-calibrated mask chain from FUN_180529fe0:
  level = am*res*scale; track += w*(level-track) (per-bin attack/release weights
  extracted from RT snapshot, rt_weights.hpp); acc = (level-track)+level; mask =
  (1/(1+K*acc))^n (K=9.8026 n=0.25966 fitted to live mask band0). min-combine.

Results (N=2048 hop=512): t1kq single-band err +0.51 dB (-14.47 vs -14.98 dB ref);
comb 4-band per-tone -3.4..+5.8 dB (old PCHIP over-cut comb ~6 dB). Adds framed_test
harness for C++ FramedDetector eval on tone1kq/comb.
 2026-08-20 13:11:27 +03:00
Matiq 5c939583f5 P4: multi-band FramedDetector (per-band twin res, sens-scaled GAIN, min-combine); tt_base 61.40%; min-combine wrong (soothe2 uses additive accumulator 0x5407c8) 2026-08-20 10:41:24 +03:00
Matiq 0cdc57972c P4: port framed_render.py model to C++ (FramedDetector: twin-res -> level -> Pchip LUT -> warp -> res^rp gain), replace empirical Detector; tt_base 49.93%->61.45% (single-band); multi-band refs still 100% (model is single-band) 2026-08-20 10:31:43 +03:00
Matiq 5a4554a9a3 P4-prep: eval_lut_bin uses float logf/expf (matches decomp FUN_180563440 logf/expf, not double std::log/exp) 2026-08-20 09:26:38 +03:00
Matiq feb44c3802 P3.4: transcribe plugin's own vectorized ln(x) (vlog.cpp, minimax ln(1+x) poly + ln2 range-reduction); correct prior misread — dispatch reaches runtime ln, not FFT; vlog_check ALL OK (rel 2.4e-7) 2026-08-20 07:12:56 +03:00
Matiq 88422034f5 P3.1: extract FFT code from rt snap (extract_fft.py); decode cplx_mul 0x8440 = in-place elementwise double mult, twiddle loader 0x39b00 stride copy, dispatcher 0x535a70->0x140a10/70->jumptable[0x1826159a0]=4; implement cplx_mul_scalar_inplace 2026-08-20 02:25:59 +03:00
Matiq 22e4599e0a P2: decode mask-accumulator combine kernels from raw bytes — combine3(0x8d60)=sub, acc_add(0x5a20)=dst+=src, acc_fma(0x3c40)=dst+=a·b; implement + levelpath_check ALL OK 2026-08-20 01:30:33 +03:00
Matiq d3db772121 P2.5: decode twin-mask factory (FUN_18056e3e0 = constant fill 2π/(count·SR)) + band LUT apply (FUN_180563a60: level->gain t^γ/power-law); implement in levelpath.cpp, levelpath_check ALL OK 2026-08-20 01:23:22 +03:00
Matiq a49e35dc25 P1.5: embed live level-tracker A[] + mask scalars into C++ (dsp/leveltrack_data.hpp); leveltrack_check sanity confirms SR=48000, A_ATTACK plateaus 0.692, attack/release=1.0 2026-08-20 01:12:12 +03:00
Matiq 433a0026e6 P2: level-tracker UPDATE-loop located (inline bidirectional IIR in FUN_180529fe0), structural module leveltrack (iir_first_order/bidirectional/unrolled4); scalar==unrolled verified; remaining = live A[] coeffs 2026-08-19 23:53:01 +03:00
Matiq 8b8895f68a P1.3: integrate FFT-conv stage (build_fir_from_window step5 memcpy + fir_from_mask + overlap-save conv) with captured WIN_WINDOW; fftconv_check confirms FIR=window[2048:4096]={0.8->1.0} 2026-08-19 23:25:21 +03:00
Matiq fd9b1fd001 P1.1: embed runtime-captured DSP tables (WIN_WINDOW, FREQAXIS@48k, WA/WB/WC/WD), dsp_ctx registry mirror, tables_check ALL OK 2026-08-19 22:48:32 +03:00
Matiq 83af30d7c0 P0: track twin.cpp/twin.hpp/rotor_kernel.hpp (were shadowed by .gitignore) 2026-08-19 22:00:22 +03:00
Matiq f8b91e8015 P0.3-4: harness reads flat params (in/out/[conf]), byte-verified trim; verify_bit_exact.py uses SOURCE WAVE+RENDER_FILE from RPP, sample-report mono/stereo 2026-08-19 22:00:14 +03:00
Matiq 58164f2952 P0: fix build (levelpath in CMake, FLOOR_LIN, exact LUT curve formulas), robust RPP param decoder (607 rpp ok), preserve f_52b570/f529fe0 disasms 2026-08-19 21:54:26 +03:00
Matiq 627e8373e6 res_power breakthrough: 500Hz residual solved (q0.1 err +0.00), decomp inventory, FUN_180563440 decoded 2026-08-19 09:56:19 +03:00
Matiq 111d47530e dsp: freqpath - warp 0x5406a8 = 0.87*x/(1+x/K) (K=exp(2.0723), x=f/2000) из FUN_180530850 (0x530900) + self-check (rel err 1.5e-07, PASS) 2026-08-18 17:08:42 +03:00
Matiq 0a4c0c959b dsp: CMake - twin.cpp/phase_table.cpp в либу, executables harness+twin_check, фикс путей dsp/ 2026-08-18 16:58:00 +03:00
Matiq b1b4f2bdf7 roadmap: detector decrypted in soothe_mem.bin, SNR 19.8dB; integrate twiddle loader 
- fft.cpp: build_twiddle via soothe::twiddle_load (Cody-Waite sin/cos); drop dup init_plan
- fft_stage.cpp: cplx_mul/stage_complex/stage_double kernels -> phase fixed (corr +0.995)
- detect.cpp: level-dependent regional floor (no bell-boost), mask 10^(-1.041*depth*floor/20)
- burst500 metrics: ref -26.07 / ours -26.13 dBFS, corr 0.99475, SNR 19.80 dB, diff -0.065 dB
- KEY: FUN_180535880/536f90 bodies are decrypted real SSE in soothe_mem.bin; dispatch
  table 0x182616008[0]=idx=4 -> 0x180009860 -> FUN_180040d40; region 0x18004xxxx = full
  detector algorithm, absent from prior fun_map/decomp (Ghidra ran on encrypted file)
 2026-08-17 20:02:32 +03:00
Matiq 847725f5fc dsp/: fix FFT inverse, WOLA normalization, detector model, twiddle loader 
- Fixed execute_inverse: removed conj bug, now uses positive twiddle only
- Added WOLA normalization factor (wola_sum/hop_ for Hann+hop=N/4)
- New detector model: floor(level) + bell_curve * boost, calibrated from
  measured data (summary.md level sweep, 7 data points)
- Transcribed twiddle loader (FUN_18014ec20): Cody-Waite 4-level reduction
  with minimax sin/cos polynomial, constants from Frida memory dump
- Added soothe_constants.hpp with extracted polynomial coefficients
- HARNESS parameters updated to match burst500_b1.rpp (depth=0.864)

Results: burst500.wav reduction now -5.8 dB vs Ref -6.8 dB (was -14.4 dB)
 2026-08-17 18:39:50 +03:00
Matiq 71870129c2 dsp/: add spectral detector with envelope + peak suppression 
- Detector computes smoothed spectral envelope
- Finds peaks exceeding envelope
- Creates per-bin suppression mask
- Applies mask in frequency domain before ISTFT
- Default params: sharpness=1.0, selectivity=0.5, depth=0.3

Verified: burst500.wav → output RMS reduced from 0.0678 to 0.0476
 2026-08-17 15:49:48 +03:00
Matiq 7dcbcf49b4 dsp/: C++ skeleton with working FFT/WOLA STFT 
- Cooley-Tukey radix-2 FFT (forward + inverse with /N normalization)
- WOLA STFT/ISTFT with Hann window (nfft=2048, hop=512)
- WAV16 read + WAV24 write (fixed aliasing bug in read)
- Fixed in-place processing bug (separate input/output buffers)
- Biquad filter (peak/shelf/reject)
- Peak detector skeleton
- MS encode/decode (M8 stereo)
- Harness: WAV16 → STFT → WAV24 passthrough verified non-zero output

Verified: burst500.wav passthrough produces output RMS=0.0678
 2026-08-17 11:25:35 +03:00