----------------------------- **While I generate an answer, here are some sources that might be helpful**: * [Update Assistant]() * [Get Assistant]() * [List Assistants]() * [Create Assistant]() -----------------------------

### Answer It sounds like you're experiencing an issue where the AI does not respond and the call is terminated due to silence. Here are a few things to check and consider: 1. **Silence Timeout**: The standard silence timeout is 30 seconds, which means if no speech is detected from the user's end within this period, the call will automatically end [(1)]() . You might want to verify if there are any issues with the microphone or the audio input on the user's end that could be causing no audio to be detected. 2. **First Message Mode**: Ensure that the

firstMessageMode

is set appropriately. If it's set to

assistant-waits-for-user

, the assistant will not speak until it hears the user. This could potentially lead to a timeout if the user does not speak [(2)]() . 3. **Voicemail Detection**: Turning off voicemail detection is appropriate if you do not want the call to be forwarded to voicemail, but this will not affect how the AI responds during the call. If the issue is that the AI is not responding at all, it might be related to other configuration settings or the specific scenario programmed into the AI [(1)]() . 4. **Check Logs for Errors**: Review the call logs for any errors or unusual activity that might indicate what is causing the AI to not respond. Look for any patterns or repeated errors . If these checks do not resolve the issue, it might be beneficial to look into the specific configurations of your assistant's settings or consult detailed logs to diagnose the problem further. Please double-check these aspects and let me know if the issue persists!

Got it. Tagging @User for additional help.

Really hoping we can fix this once and for all this issue is killing me.

and on the third try it works. i dont get it .

this looks like a transcriber issue

no transcript for both of these calls

another one 0093e9c1-cb24-4e52-854c-5f434ad74598

this one has a transcript

Also note usually i run in to issues when testing things at weird hours of the day or on what would be considering slow periods. Not sure if some sort of scaling issue on teh vapi end where services go to sleep when not in use or something .

is this related to deepgram. ?

Orginally i thought it was transcription but ive had another call where it did the same thing but transcirption was there so think the issue is something diff.

If yes remove it. It should work after that I guess

Yup.

@Sahil look at this one

Can you try with different transcriber?

this is a new one.

but yeah trying talkscriber now

Sahil is this something that will be fixed, not really happy with the whisper model

Let me check once again.

seems each of my failure scenarios are all different

sometimes the there is no transcription adn no response, some times there is a transcription and no response, and sometimes it takes 20-30 seocnds to respond. And 70% of the time it works fine .

could it be specific to using nova-2-phonecall vs nova-2. ?

nova-2-phonecall doesn’t support any other language apart from en

yeah our calls are in en

the keywords work in nova-2

so i guess for now to mitigate problems will use nova-2 , nova-2-phonecall is lil better in performance for our use case but i think nova-2 will be fine def better than whisper

so i dont know if its a fluke but i had the 20 secodn delay issue on two calls then switched to nova-2 and all is working fine now. The issue is always sporatic so cant be sure if it's fixed or if just a fluke but i will report back if it continues

Well it was working fine then it stopped looks like we are hard down right now @Sahil

"Sorry an application error has occured"

the default twilio error message

An application error has occurred

Sometimes it works sometimes it doesn’t

Like 50% of my calls get the Twilio error

Think the vapi backend is throwing errors