ASR + Smart Turn Live Test

Simulates the telephone path: mic captured as 8 kHz PCM Int16 mono (the rate a real phone line delivers), then upsampled to 16 kHz before being sent to the streaming ASR and Smart Turn services. Audio streams continuously; ASR end events drive Smart Turn analyze. Each Smart Turn probability is fused with six language-agnostic ASR signals — excess trailing silence, tail-confidence drop, fragment size, speech density, cross-segment cadence, and turn duration — into a final commit probability. See docs/smart_turn/SIGNALS.md.

ASR Service

Smart Turn Service

Enable Smart Turn for this session Captured at Start. ASR is always tested. When off, the session does not init the Smart Turn WS and skips the analyze call on ASR end. Locked while a session is running.

Smart Turn WebSocket URL

Token (optional)

Confidence threshold (client-side)

Advanced Smart Turn params

stop_secs

max_duration_secs

rms_threshold

Fusion coefficients — applied to Smart Turn prob at each ASR end (see docs/smart_turn/SIGNALS.md)

gap divisor

gap max

tail Δ threshold

tail conf max

tail penalty

fragment max words

fragment conf max

fragment penalty

density threshold

density penalty

cadence Δt max (s)

cadence penalty

turn start (s)

turn ramp (/s)

turn max

ST trust cutoff

last-word conf cutoff