Commit Graph

5 Commits

Author SHA1 Message Date
kami 62cc072f8c Add golden-audio STT tests against real whisper.cpp (#288)
Four committed WAV fixtures go through the real whisper.cpp binding in
cmd/mavsttd, so a wrong model, a wrong language hint, a broken resample
or a regressed silence gate fails `make test` instead of surfacing as
Maven mishearing him.

The fixtures are piper-synthesised, not recorded: scripts/gen-stt-fixtures.sh
drives the vendored piper with the ru_RU-irina voice Maven already speaks
with, so nothing of the owner's voice is committed and every fixture is
reproducible. 360K total for three Russian clips and one English.

Matching is tolerant on purpose. Golden transcripts move with the model,
so each case asserts intent-carrying keywords (prefix match, so Russian
inflection does not fail it) plus a word error rate ceiling, not an exact
string. The matcher is unit-tested on its own and needs no model.

TestGoldenAudioTranscription skips when models/stt/ggml-small.bin is
absent, so `make test` still passes on a box without models.
TestGoldenFixturesAreCanonical runs everywhere and checks the WAVs are
16k mono s16le and would clear mavsttd's own silence gate.
2026-08-01 05:32:10 +04:00
kami d52f60c54e maven: fix test mocks for CalendarEvents interface (verification)
- Add CalendarEvents method to recordingAPI in auth_test.go
- Add CalendarEvents method to fakeCore in handlers_test.go

Co-Authored-By: opencode <opencode@anthropic.com>
2026-07-06 04:20:16 +04:00
kami 2740c2f685 mavsttd: silence gate — don't feed non-speech to whisper
Whisper hallucinates subtitle-credit boilerplate ("Редактор субтитров …") on
silence/room-noise, which then got stored as tap:voice facts. Gate before the
model: drop clips shorter than -min-ms (default 300) or below -silence-rms
(default 0.01 normalized RMS). Both are flags — the mic floor is hardware
specific. Returns empty transcript (same as whisper's no-segments path), so
nothing downstream changes.

gateReason is pure and unit-tested (silence/short/quiet → dropped, loud+long →
passes). ponytail: energy gate, not a real VAD; upgrade to WebRTC VAD or
whisper no_speech_prob if too blunt.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-07-04 00:02:44 +04:00
kami b77f209686 voice: pre-route quiet-hours toggle, whisper ctx cancellation, stale reply fix
- resolveQuietToggle runs in HandlePushToTalk before the router so
  'тихий режим' works regardless of classifier confidence.
- whisper_full() runs in a goroutine with ctx.Done() select so the
  handler returns promptly on timeout/shutdown.
- StubReplier.IntentQuery no longer claims query is unimplemented.
2026-07-03 10:56:44 +02:00
kami e00cb07658 fix zombie leak, add quiet-hours toggle, improve query reply, configurable router threshold, JS dashboard 2026-07-03 00:42:35 +02:00