Hi, I am using a built in Vapi voice ( Kylie ) in production and in some calls it breaks half sentence and stops and the customer needs to nudge the model so it continued e.g "Hello?". We are running thousands of calls a month on your platform and ant to increase that. The only thing why we chose you is this voice. We need this fixed ASAP.

To help prevent the Kylie voice from breaking mid-sentence, try making your

stopSpeakingPlan

less sensitive by increasing the

voiceSeconds

value. For example:

json
"stopSpeakingPlan": {
  "numWords": 0,
  "voiceSeconds": 0.4,  // More conservative, reduces false positives
  "backoffSeconds": 1.0
}

Apply this in your voice pipeline configuration to reduce premature interruptions and improve conversation flow. For more details, see the [Vapi voice pipeline configuration guide](https://docs.vapi.ai/customization/voice-pipeline-configuration). Source: - [Voice pipeline configuration](https://docs.vapi.ai/customization/voice-pipeline-configuration)

Hi there, Thank you for your message. Our team is currently out of the office. We operate Monday through Friday, from 9:00 AM to 8:00 PM Pacific Standard Time (PST). We’ll get back to you as soon as possible during our normal business hours. If your message is urgent, please mark it accordingly or include “URGENT” in the subject line, and we’ll do our best to respond promptly. Warm regards, Vapi Customer Support Team

It is not stop speaking plan, it breaks in complete silence mid sentence

What is the max tokens value? I faced a similar problem because I set a low value (I didn't want my agent to speak too much).

Try setting your assistant temperature to 0.4

Hi @kyle I'm having the same issue here. Why would temperature to 0.4 make a difference in this case - what is the thinking behind it (and what is actually causing the issue)? A recent (just now) example is call ID

019b0992-66c7-7bba-a18e-415da979fdb5

from Org ID

1cbb0dbc-cacb-411e-bbbc-bc459102d14b

. No error logs (or even warnings), the LLM seems to have sent back the full message in this step, no logs indicating this is an issue with the

stopSpeakingPlan

, and no other errors on the account.