Primary users are non-technical people who want Whisper without docs or
ML setup; technical users must keep full control as if running it on
their own device.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Typecheck: typescript 7 devDependency, `bun run typecheck`; CI gains a
`check` stage (install, typecheck, fast tests) gating image builds (D-038).
- Light theme: every colour is a token; prefers-color-scheme: light
overrides them. Inter Variable self-hosted from @fontsource-variable/inter,
kept out of Bun's CSS inlining and precached (scripts/fonts.ts, D-037).
Backend select no longer overflows on narrow phones.
- tests/live-real-model.test.ts: time budget scales with clip length and
en_alien gets a measured WER bar; it had failed since July (D-039).
- Delete the temporary isolate_wer2.ts; fix icon "not checked in" comments
and the README's model-weights location.
- Agent docs: branch naming, typecheck, CI, Bun CSS gotcha, permissions.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Root AGENTS.md is now the single source of truth for coding agents;
CLAUDE.md, GEMINI.md and docs/CLAUDE.md are pointers. The product spec
moves from docs/CLAUDE.md to PRODUCT.md, DESIGN.md records the shipped
visual system, and docs/PERMISSIONS.md defines agent permission tiers
rendered to .claude/, .codex/ and .gemini/. Adds docs/specs/ and
docs/plans/, logs issues noticed (not fixed) in WorkBreakdown.md.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Rename the app from "AU Whisper Web" to "Whisper Web" in the README
title and app header, rename the package to whisper_web, and drop the
au_ prefix from the transformers patch comment and the scratch path in
isolate_wer2.ts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Measured with the ort-web-under-Bun harness: no reliable speed difference on
clean, language-matched audio. On out-of-distribution audio (wrong/undetected
language), the multilingual model's extra forced task token measurably
changes how long its repetition-loop hallucination runs before EOS (434 vs 67
tokens on the da_festuge fixture), which is where the reported slowness comes
from. Both models resolve to identical, merged, KV-cache-enabled decoders at
the same dtype — no catalog/config fix applies here. Documents the mechanism
and flags a future no_repeat_ngram_size/repetition_penalty mitigation as the
actionable follow-up if this recurs.
startLive()'s browser-engine branch flipped liveStatus to "listening" (the
red live-dot, "Listening..." hint) the instant the mic was granted, even if
the Whisper pipeline was still loading — the user saw an active-looking
session while nothing was being transcribed yet. transcribeLiveSegment()
already correctly awaits the model-ready gate before posting a segment, so
this was purely a misleading-status bug.
Adds a "loading-model" LiveStatus between "starting" and "listening";
startLive() now waits on the same browserModelReady() gate the transcribe
path uses before claiming "listening", with race-safe cleanup if the load
fails while waiting. Server-engine live is unaffected.
Adds Decisions.md D-034; updates tests/live.md's manual verification steps.
Transformers.js's progress_callback always fires status:"download" regardless
of whether the file is actually served from the Cache API — there's no
cacheHit signal for the worker to forward. Since App.tsx already tracks which
models are downloaded (for ModelPicker's badges), it now checks that before
starting a load and threads a loadedFromCache flag into ModelLoadProgress, so
the heading/hint reflect reality for the common already-cached case.
Adds Decisions.md D-033.
