Record the batched-VAD measurements and check heartbeat, bursts and cancel against the real sidecar

Co-Authored-By: Claude Opus 5.5 <[email protected]>

docs/Architecture.md

@@ -204,7 +204,9 @@ environment). See `WorkBreakdown.md` for the itemized checklist.)
    (`server-py/`) that Bun spawns and supervises (D-013; faster-whisper has no
    browser/WASM build, so it is server-only — the browser stays on Transformers.js).
    `src/server/transcribe.ts` bridges the sidecar's segment stream to the streamed
-   JSON-lines HTTP response (D-011). The shared catalog maps model ids to
+   JSON-lines HTTP response (D-011). File requests (`mode: "file"`) decode with
+   faster-whisper's batched pipeline over Silero-VAD speech chunks; server-live
+   passes send `mode: "live"` and stay sequential (D-WS5a). The shared catalog maps model ids to
    `Systran/faster-whisper-*` CTranslate2 conversions. If the sidecar is absent,
    the server engine reports unavailable; the browser/anonymous flow is unaffected.
 4. Both engines implement the same `Engine` interface in `src/lib/engine.ts`, so the

docs/Decisions.md

@@ -1224,3 +1224,48 @@ whole-file output is equally bad, which is exactly what the absolute bar
 exists to catch.
 **Consequence:** full `bun test` ≈ 19 min on CPU (en_alien ≈ 20 min of it
 on a cold run); the fast suite excludes this file.
+
+## D-WS5a — Server engine: batched faster-whisper + tuned Silero VAD for files; live passes stay sequential
+**Status:** Accepted (2026-10-30, spec 001 Goal 5)
+`server-py/sidecar.py` decoded every request with the sequential
+`WhisperModel.transcribe`, no VAD.
+**Decision:** the sidecar request gains `mode: "file" | "live"` (absent ⇒
+`"file"`, so an older Bun still gets the new path). `"file"` (`POST
+/api/transcribe`) runs `BatchedInferencePipeline(model).transcribe(audio,
+vad_filter=True, vad_parameters={threshold: 0.2, speech_pad_ms: 800},
+batch_size=8)`; the pipeline is a weightless wrapper cached and rebuilt with
+the one resident model. `"live"` (server-live passes, `src/server/live.ts`)
+keeps the sequential decoder. Beam size stays at the default 5.
+**Measurements** (CPU int8, AMD Ryzen 9 9950X, 16 cores shared with other
+jobs — load average 7–24 during runs, so wall-clock is ±20 %; WER via jiwer
+on the `tests/live-real-model.test.ts` normalisation vs `tests/audio_files/*.txt`):
+
+| model | audio | config | wall-clock | WER |
+|---|---|---|---|---|
+| large-v3-turbo | en_alien (888 s) | sequential (before) | 97 s / 130 s (two runs) | 0.097 / 0.067 |
+| large-v3-turbo | en_alien | sequential + VAD | 79 s | 0.136 |
+| large-v3-turbo | en_alien | batched, default VAD (0.5 / 400 ms) | 47 s / 59 s | 0.117 |
+| large-v3-turbo | en_alien | batched, beam 1 | 44 s | 0.114 |
+| large-v3-turbo | en_alien | **batched, VAD 0.2 / 800 ms (after)** | **51 s / 58 s** | **0.086** |
+| large-v3-turbo | en_alien | batched, VAD 0.2 / 800 ms, word-level timestamps | 56 s | 0.144 |
+| large-v3-turbo | da_festuge (21.5 s, da) | sequential (before) | 2.9 s | 0.094 |
+| large-v3-turbo | da_festuge | batched, default VAD | 2.9 s | 0.141 |
+| large-v3-turbo | da_festuge | **batched, VAD 0.2 / 800 ms (after)** | **2.7 s** | **0.109** |
+| medium.en | en_alien | sequential (before) | 114 s | 0.126 |
+| medium.en | en_alien | **batched, VAD 0.2 / 800 ms (after)** | **67 s** | **0.111** |
+
+The sequential decoder is not deterministic run to run on en_alien (temperature
+fallback; 0.067–0.097 over three runs), while the batched result repeated
+exactly. Default-VAD errors were deletions (121 vs 64 sequential): Silero at
+0.5 dropped quiet speech under music; 0.2 / 800 ms recovered most of them.
+da_festuge's gap is one word of 64. Live-style 6 s passes over the first
+180 s of en_alien: sequential 2.45 s/pass, batched 2.12, sequential + VAD
+1.88 — within the load noise, each losing a few words, and real live
+passes are already speech-gated client-side (D-016), so live stays as it was.
+**Consequences:** file transcriptions are ≈ 1.7–2.5× faster on CPU at
+equal-or-better WER on en_alien (one word worse on the 21 s Danish clip). Segments are coarser — one per VAD chunk (≤ 30 s) instead of one per
+sentence — because word-level timestamps (`without_timestamps=False`) cost
+WER here (0.144); SRT/VTT cues from the server engine are correspondingly
+long. Segments arrive in bursts of 8 chunks (≈ 4 min of speech) and the first
+burst can take ~15–25 s on a long file; the D-021 heartbeat covers it, and
+cancel lands at the next burst boundary. No new env vars.

docs/WorkBreakdown.md

@@ -145,6 +145,9 @@ Notes (M4):
       (CTranslate2) via a Python sidecar in `server-py/`, spawned/supervised by
       Bun (D-013 supersedes the Transformers.js-under-Bun plan); map the shared
       model catalog to faster-whisper model ids in `src/lib/models.ts`
+- [x] Batched inference + tuned Silero VAD for server-engine file
+      transcriptions, live passes sequential (spec 001 Goal 5, D-WS5a):
+      ≈ 2× faster on CPU int8 at equal-or-better WER; segments ≤ 30 s chunks
 - [x] Degrade gracefully when Python/the sidecar is absent (server engine
       unavailable; browser engine untouched — anonymous path deploys with Bun alone)
 - [x] `POST /api/transcribe` — auth-guarded, streamed JSON-lines response (D-011),