Josh L
04/27/2025, 8:54 PMVapi
04/28/2025, 4:35 PMAbe S
04/28/2025, 7:28 PMJosh L
04/28/2025, 11:15 PMVapi
04/30/2025, 2:01 PM{
"transcriber": {
"model": "nova-2",
"language": "en",
"provider": "deepgram",
"endpointing": 300
},
"startSpeakingPlan": {
"waitSeconds": 0.4,
"smartEndpointingPlan": {
"provider": "livekit",
"waitFunction": "20 + 500 * sqrt(x) + 2500 * x^3"
}
}
}
Hey Josh, could you try two to three test calls with the following changes:
1. Transcriber end pointing set to 300ms
2. Start speaking plan smart endpointing plan live kit wait function change to an updated value
What's happening is the Endpointing model is waiting a long time before considering the user has finished speaking.Vapi
04/30/2025, 2:02 PMJosh L
04/30/2025, 6:42 PMVapi
04/30/2025, 8:34 PMJosh L
05/04/2025, 1:04 AMJosh L
05/04/2025, 1:04 AMVapi
05/05/2025, 11:03 PM