8 Commits

Include renames

Author	SHA1	Message	Date
mathiasandClaude Sonnet 5.5	a2027c9f9b	chore(fixtures): invented text in place of words that may come from the private recording; the research note no longer names the private file (adversarial review F10) ... Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>	2026-10-02 16:49:54 +02:00
mathiasandClaude Sonnet 5.5	2e9d76dad0	docs(research): first blind check of two-engine reconcile on a real recording (aggregates only) ... Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>	2026-10-02 08:01:31 +02:00
mathiasandClaude Sonnet 5.5	e3aa661cf2	docs(research): Low Power Mode confirmed as the cause of slow gemma4-31b (22.3 vs 7.9 tok/s decode) ... Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>	2026-10-02 07:13:10 +02:00
mathiasandClaude Sonnet 5.5	e182a6b07d	docs(research): can a local model adjudicate two ASR transcripts? ... Eleven local models, 120 calls each, scored against the speaker's blind answers on 30 disagreements, with pass criteria fixed before the run. gemma4-31b (97%) and gemma4-26b (92%) pass; nothing else clears accuracy and order-consistency. Records why 31b is slow (MLX gives the same speed; Low Power Mode is the unproven suspect) and the caveats: 30 items, one recording, thinking models under-budgeted. Aggregate numbers only; this repo is public. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>	2026-10-01 23:14:27 +02:00
mathiasandClaude Sonnet 5.5	ebc6cc3ff8	docs(research): kb-whisper vs pianissimo on one real hour of Swedish speech ... A near tie on accuracy (blind A/B by the speaker: 17 vs 13 on the 30 spans that change meaning) and a 6x difference in speed (40 s vs 4m08s for 63.7 min). Different error profiles: pianissimo better on terms and numbers, kb-whisper on everyday words. Aggregate numbers only -- no transcript text or domain terms, this repo is public. Also records the mid-word wrapping bug the comparison found and its fix. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>	2026-10-01 19:45:56 +02:00
mathiasandClaude Sonnet 5.5	4d3e2a471a	docs(research): chat over a long transcript -- what the offered models really hold ... phi4-mini 400s at ~13k tokens; gemma4-26b is correct to 90,000 chars and silently loses the HEAD at 120,000 (no error, no truncation flag). The end-needle test alone passed at 120k -- only a head-needle exposed it. Records the default-model trap and the iguana contention hazard. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>	2026-10-01 17:08:13 +02:00
mathiasandClaude Sonnet 5.5	feb108def3	docs(research): measure the audio length ceiling before designing chunking (#33) ... No transcription ceiling found up to 90 min: one 174 MB request through the LiteLLM NodePort transcribed in 440 s. The first real failure is the public gateway ingress rejecting bodies over ~8-10 MB with 413, not whisper or a Metal OOM. Also records that srt arrives wrapped in a JSON envelope, that the two backends differ in timestamp granularity, and that the existing chat path truncates at 18k chars (~22 min of speech). Synthetic Swedish TTS fixture only; no real transcript. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>	2026-10-01 16:33:34 +02:00
mathiasandClaude Opus 4.8	92da25d01e	docs(research): Vimeo + Whisper transcript feasibility findings ... Vimeo texttracks API is owner-only (worse than YouTube, no public timedtext) → defer. Whisper viable as a no-caption fallback; berget/whisper-large-v3 already on the gateway (cloud), local options on iguana/koala carry setup+GPU-contention cost (ADR-007). Both investigated inline (session sub-agents are network-sandboxed). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>	2026-06-02 23:14:03 +02:00