Why am I getting

skipping user input, speech scheduling is paused

?

Thanks for reaching out, Daniel. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

I don't get where this is coming from? The voice agent greets me and then I want to say something. However, my speech is not processed for some reason

To get to the bottom of this, could you share a bit more about your setup? For example: • Are you using a custom

on_enter

method for the agent's greeting? • Which turn detection mode are you working with (like VAD, STT, or manual)? • Does this happen with just a simple greeting, or only when the greeting logic gets more complex? • Have you noticed any specific error messages or state change events in your logs? Sources: examples/voice_agents/speaker_id_multi_speaker.py | pipeline-tts/only_greet.py | How does end-of-utterance detection work in conversations?

Yes. I do:

async def on_enter(self):
        await self.session.generate_reply()

I am using VAD It always happens

Using a custom

on_enter

method with

session.generate_reply()

is actually pretty common and shouldn’t cause any issues with the speech scheduling system. The speech created by

generate_reply()

goes through the usual scheduling process, so the timing and handling should work smoothly. Since you’re using VAD and this issue keeps popping up, it might have something to do with how the system switches from the greeting to the listening state. Could you check if the agent’s state changes to "listening" right after the greeting? Also, are there any specific log messages or warnings related to VAD or turn detection that you’ve noticed? Sources: livekit-agents/livekit/agents/voice/agent_activity.py | telephony/answer_call.py | main.py