In some cases, my Voice Agent does not reply, and I have noticed that this happens whenever there is no transcription received. I have already lowered the Confidence Threshold from 0.5 to 0.1, which helped a bit. I also tested with the Confidence Threshold set to 0, but I still experience situations where no transcription is generated.
If there is no transcription within 3 seconds after the user speaks, I would like my Voice Agent to say something like: “Sorry, I didn’t catch that. Could you please repeat that?”
I have already added an instruction to my system prompt: “Whenever there is no, unclear, or out-of-context transcription, always ask again in a friendly and relaxed way, for example: ‘Sorry, could you please repeat that?’ This also applies if the answer does not make sense or does not match the question.”
However, this does not work. When there is no transcription, the Voice Agent simply remains silent and then closes the call once the Silence Timeout is reached.
Any ideas?