Hello!
I am using faster-whisped in a container. It serves for STT in HA assist. I use to talk to it via the companion app installed on my android phone. The problem is that when I open assist and start talking to it, my first addressing always ends with VAD timeout:, i.e. it does not detect the end of speech:
VAD end of speech detection timed out after 15.0 seconds
In the containers logs this appears:
DEBUG:wyoming_faster_whisper.handler:Language set to en
DEBUG:wyoming_faster_whisper.handler:Audio stopped. Transcribing with initial prompt=None
INFO:faster_whisper:Processing audio with duration 00:15.000
DEBUG:faster_whisper:Processing segment at 00:00.000
INFO:wyoming_faster_whisper.handler: Hello?
DEBUG:wyoming_faster_whisper.handler:Completed request
the consequent request are processed without significant delay, so it is the problem of conversation start only.
This container serves 3 of the assistans I have made for different languages if this info is usefull.
Hello!
I am using faster-whisped in a container. It serves for STT in HA assist. I use to talk to it via the companion app installed on my android phone. The problem is that when I open assist and start talking to it, my first addressing always ends with VAD timeout:, i.e. it does not detect the end of speech:
In the containers logs this appears:
the consequent request are processed without significant delay, so it is the problem of conversation start only.
This container serves 3 of the assistans I have made for different languages if this info is usefull.