tocmo0nlord/avc-phone-ai
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Cut smart-turn INCOMPLETE wait 3s -> 1.5s (SMART_TURN_STOP_SECS)

Browse Source 
Follow-up test call: no more cancelled replies, but 3-5s response gaps on
turns the smart-turn model judged INCOMPLETE ("I'm due to my annual exam.") -
it waited the library-default 3s of silence before triggering the LLM. Build
the stop strategy explicitly with SmartTurnParams(stop_secs=1.5), env-tunable.
A caller who really does resume just yields a follow-up turn, which is safe
now that interruption broadcasts are off.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
tocmo0nlord
2026-07-04 03:23:03 +00:00
parent 3ed63d8ea9
commit 94e2ca1902
3 changed files with 39 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
.env.example
View File

@@ -68,6 +68,9 @@ VAD_CONFIDENCE=0.5
VAD_MIN_VOLUME=0.15
VAD_START_SECS=0.1
VAD_STOP_SECS=0.5
# Extra silence the smart-turn analyzer waits when it judges an utterance INCOMPLETE
# before ending the turn anyway (library default 3s caused ~3.5s dead air on some turns).
#SMART_TURN_STOP_SECS=1.5
# Deterministic slot memory (callstate.py): injects an ALREADY-COLLECTED / STILL-NEEDED
# checklist into the system prompt each turn + merges VAD-fragmented user turns, so the
# local 8B stops re-asking for name/reason/phone. Default: on for ollama, off for anthropic.