Hello Team,
We have been testing our AI solution for a while now and it freezes a lot during conversation, sometimes it even takes minutes for it to create the response. We use OpenAIs Realtime model. I would assume it has something to do with the turn detection settings we have configured. Has anyone experienced or solved this issue beforehand? This is the current configuration for turn detection:
turn_detection=TurnDetection(
type="semantic_vad",
eagerness="low",
create_response=True,
interrupt_response=True,
),