linnet/app/services/annotator.py
pyr0ball 7e14f9135e feat: Notation v0.1.x scaffold — full backend + frontend + tests
FastAPI backend (port 8522):
- Session lifecycle: POST /session/start, DELETE /session/{id}/end, GET /session/{id}
- SSE stream: GET /session/{id}/stream — per-subscriber asyncio.Queue fan-out, 15s heartbeat
- History: GET /session/{id}/history with min_confidence + limit filters
- Audio: WS /session/{id}/audio — binary PCM ingestion stub (real inference in v0.2.x)
- Export: GET /session/{id}/export — downloadable JSON session log
- ContextClassifier background task per session (CF_VOICE_MOCK=1 in dev)
- ToneEvent SSE wire format per cf-core#40 (locked field names)
- Tier gate: CFG-LNNT- prefix check, 402 for paid features

Vue 3 frontend (port 8521, Vite + UnoCSS + Pinia):
- NowPanel: affect-aware border tint, subtext, prosody flags, shift indicator
- HistoryStrip: horizontal scroll, last 8 events with affect color
- ComposeBar: start/stop session, SSE connection lifecycle
- useToneStream: EventSource composable
- useAudioCapture: AudioWorklet → Int16 PCM → WebSocket (v0.1.x stub)
- audio-processor.js: 100ms chunk accumulator in AudioWorklet thread
- Respects prefers-reduced-motion globally

26 tests passing, manage.sh, Dockerfile, compose.yml included.
2026-04-06 18:23:52 -07:00

23 lines
902 B
Python

# app/services/annotator.py — VoiceFrame → ToneEvent pipeline
#
# BSL 1.1: this file applies the cf-voice inference results to produce
# session-scoped ToneEvents. The actual inference runs in cf-voice.
from __future__ import annotations
from cf_voice.models import VoiceFrame
from app.models.tone_event import ToneEvent
def annotate(frame: VoiceFrame, session_id: str, elcor: bool = False) -> ToneEvent | None:
"""
Convert a VoiceFrame into a session ToneEvent.
Returns None if the frame is below the reliability threshold (confidence
too low to annotate confidently). Callers should skip None results.
elcor=True switches subtext to Mass Effect Elcor prefix format.
This is an easter egg -- do not pass elcor=True by default.
"""
if not frame.is_reliable():
return None
return ToneEvent.from_voice_frame(frame, session_id=session_id, elcor=elcor)