Kazeia-engine/dist/jni
Richard Loyer c332943181 Decoder chraac en sous-process : -20% decoder, -9% pipeline, talker intact
Architecture pour exploiter le gain chraac sans casser le talker :
  - kazeia_decoder_chraac : binaire static contre libqwen3tts-decoder.a chraac
    + ggml chraac statiques (9.8 MB autonome, pas de dép .so chraac).
    Args : <decoder.gguf> <codes.bin> <T> <out.wav>. Lit codes int32 time-major,
    transpose codebook-major, forward, écrit WAV.
  - tts_engine : si KZTTS_DECODER_SUBPROC=path + KZTTS_DECODER_GGUF=path posés,
    fork+execl au lieu de dec.forward intégré. Codes via /data/local/tmp tmpfile,
    WAV via rename. Le talker+CP continuent sur ql/ggml (qualcomm) qui marche.

Mesures Pad3 KZTTS_THREADS=6 GGML_NUM_THREADS=8 seed=42 :

  Phrase 'Bonjour Kazeia' (33 frames) :
    A baseline qualcomm : decoder 3.29s, total 6.61s, RTF 2.40
    B chraac subproc    : decoder 2.63s, total 6.06s, RTF 2.20
    -> decoder -20%, total -8.3%, audio RMS 1743 identique

  Phrase 'Bonsoir, comment tu te sens ce soir ?' (41 frames) :
    A baseline : decoder 3.95s, total 8.16s, RTF 2.39
    B subproc  : decoder 3.15s, total 7.38s, RTF 2.16
    -> decoder -20%, total -9.6%

Audio cohérent au baseline (range, RMS quasi identiques), imperceptible
à l'oreille (cf cas 'Bonsoir' envoyé).

Le gain chraac historique sur decoder seul est désormais accessible SANS
casser le talker.

Code committé opt-in (KZTTS_DECODER_SUBPROC absent par défaut = baseline intact).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-30 18:14:09 +02:00
..
EngineLlmEngine.kt chantier B TTS #1: API JNI embeds-only + M-RoPE qwen3 dense 2026-05-28 13:37:17 +02:00
TtsEngine.kt chantier B TTS #11: libkazeia_tts.so + Kotlin wrapper, JNI validé bout-en-bout 2026-05-28 21:54:21 +02:00
build_prefill.cpp chantier B TTS #4: P3.1+P3.2 = x_vector + prefill_embeds depuis input_ids (bit-exact) 2026-05-28 15:05:23 +02:00
cp_inference.cpp T2.1 CP heads NEON 16-way: CP -33%, pipeline -16%, bit-exact 2026-05-29 17:26:54 +02:00
cp_inference.h chantier F (CP HMX) : path sched livré, gain absent (overhead per-call) 2026-05-29 16:03:41 +02:00
cp_validate.cpp chantier B TTS #5: P3.3 = vrai CP runner intégré (bit-exact) 2026-05-28 15:11:07 +02:00
decoder_subprocess.cpp Decoder chraac en sous-process : -20% decoder, -9% pipeline, talker intact 2026-05-30 18:14:09 +02:00
dual_ctx.cpp dist: prefill HTP = CASSÉ (sortie charabia), prefill obligatoirement CPU 2026-05-27 09:42:59 +02:00
dual_ctx_mt.cpp dist: JNI option C câblé (prefill HTP / decode CPU), validé multi-tour 2026-05-27 10:24:55 +02:00
kazeia_engine_jni.cpp chantier B TTS #2: fix M-RoPE positions en embeds-mode + validation bit-exact 2026-05-28 13:54:03 +02:00
kazeia_text_tokenizer.cpp chantier B TTS #10: tokenizer Qwen3 BPE via llama_tokenize, texte arbitraire 2026-05-28 21:21:53 +02:00
kazeia_text_tokenizer.h chantier B TTS #10: tokenizer Qwen3 BPE via llama_tokenize, texte arbitraire 2026-05-28 21:21:53 +02:00
kazeia_tts_jni.cpp chantier B TTS #11: libkazeia_tts.so + Kotlin wrapper, JNI validé bout-en-bout 2026-05-28 21:54:21 +02:00
sampler.cpp chantier B TTS #7: sampler HF-style top_k+top_p+rep_penalty (générique) 2026-05-28 16:55:05 +02:00
sampler.h chantier B TTS #7: sampler HF-style top_k+top_p+rep_penalty (générique) 2026-05-28 16:55:05 +02:00
test_engine.c Valide device: generate() bout-en-bout OK (test_native), bridge pret integration 2026-05-24 22:05:44 +02:00
test_engine.cpp Valide device: generate() bout-en-bout OK (test_native), bridge pret integration 2026-05-24 22:05:44 +02:00
test_engine_2calls.cpp chantier B TTS #11: libkazeia_tts.so + Kotlin wrapper, JNI validé bout-en-bout 2026-05-28 21:54:21 +02:00
test_jni_native.cpp dist: multi-tour propre (generateRaw + ChatSession) + JNI validé end-to-end natif 2026-05-27 11:18:20 +02:00
test_jni_tts.cpp chantier B TTS #11: libkazeia_tts.so + Kotlin wrapper, JNI validé bout-en-bout 2026-05-28 21:54:21 +02:00
test_native.cpp v4 fix #277: ngl0 CPU decode 14tok/s, 4s/tour, usable 2026-05-25 10:18:31 +02:00
test_talker.cpp chantier B TTS #1: API JNI embeds-only + M-RoPE qwen3 dense 2026-05-28 13:37:17 +02:00
test_talker_replay.cpp chantier B TTS #2: fix M-RoPE positions en embeds-mode + validation bit-exact 2026-05-28 13:54:03 +02:00
test_tokenizer.cpp chantier B TTS #10: tokenizer Qwen3 BPE via llama_tokenize, texte arbitraire 2026-05-28 21:21:53 +02:00
tts_engine.cpp Decoder chraac en sous-process : -20% decoder, -9% pipeline, talker intact 2026-05-30 18:14:09 +02:00
tts_engine.h POC streaming TTFB ÷4-6 (KZTTS_STREAM_CHUNK=K) 2026-05-29 22:56:14 +02:00
tts_orchestrate.cpp chantier B TTS #6: P3.4+P3.5 = pipeline complet 100% tablette bout-en-bout 2026-05-28 15:25:00 +02:00
tts_pipeline.cpp POC streaming TTFB ÷4-6 (KZTTS_STREAM_CHUNK=K) 2026-05-29 22:56:14 +02:00