pyr0ball
|
6e17da9e93
|
feat: AudioEvent models, classify_chunk() for per-chunk request-response path
- events.py: AudioEvent dataclass + ToneEvent with affect, shift_magnitude,
shift_direction, prosody_flags; make_subtext() for generic/Elcor formats
- context.py: classify_chunk(audio_b64, timestamp, prior_frames, elcor)
returns list[AudioEvent]; mock mode uses MockVoiceIO RNG, real raises NotImplementedError
- ToneEvent.__post_init__ pins event_type='tone' (avoids MRO default-field ordering bug)
- Elcor mode: same classifier output, Elcor speech-prefix wording; all tiers
|
2026-04-06 16:53:10 -07:00 |
|
pyr0ball
|
35fc0a088c
|
feat: initial cf-voice stub — VoiceFrame API, mock IO, context classifier
- VoiceFrame dataclass: label, confidence, speaker_id, shift_magnitude, timestamp
- MockVoiceIO: async generator of synthetic frames on a timer (CF_VOICE_MOCK=1)
- ContextClassifier: passthrough stub wrapping VoiceIO; _enrich() hook for real classifiers
- make_io() factory: mock mode auto-detected from env, raises NotImplementedError for real audio
- cf-voice-demo CLI entry point for quick smoke-testing
- 12 tests passing; editable install via pip install -e ../cf-voice
|
2026-04-06 16:03:07 -07:00 |
|