When I set the assistant to speak first, the message starts too early once the call initiates and the first couple of letters are ignored. You listen to a word without its beginning. I tried using this ", ... . ... ." before my message, but it still does not fix it. Is there a way to make the assistant wait few milliseconds before it starts talking? Thanks!

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

To prevent the beginning of your assistant's message from being cut off, add a short delay using the

waitSeconds

parameter in your

startSpeakingPlan

. For example:

json
{
  "startSpeakingPlan": {
    "waitSeconds": 0.5
  }
}

This will make the assistant wait 500ms before speaking, helping ensure the full message is heard. Reference: See the Voice Pipeline Configuration documentation for more details ([source](https://docs.vapi.ai/customization/voice-pipeline-configuration)). Source: - [Voice Pipeline Configuration](https://docs.vapi.ai/customization/voice-pipeline-configuration)

I have tried this, it does not add any delay to the first message. I guess the delay is for the rest of the conversation.

I see what you mean this usually happens when the assistant audio starts before the call audio channel is fully ready, so the first phonemes get clipped. I’ve run into this a few times when working with voice agents and there are a couple of ways to introduce a small delay so the greeting plays cleanly. If you’d like, tell me which stack you’re using and I can help you sort it out.

Hey, try switching the firstMessage mode to

assistant-speaks-first-with-model-generated-message

or increase the waitSeconds in startSpeakingPlan:

"startSpeakingPlan": {
      "waitSeconds": 1.0
    }

Hi, the startSpeakingPlan does not work for the first message. The assistant-speaks-first-with-model-generated-message indeeds works fine, but it delays too much sometimes and the users can get confused and start speaking before the agent. @Diana precious I am using a Azure OpenAI, eleven labs (TTS) and a custom transcriber (STT). The custom STT should not affect the issue. I am using transient agents and set the firstMessage using the assistant-speaks-first mode. Thank you!

Got it that makes sense. When you’re using transient agents with assistant-speaks-first, the timing between the call connection and the TTS stream can cause that awkward delay or clipping, especially with Azure + ElevenLabs in the loop. I have dealt with similar voice-agent timing issues before, so if you want, share a bit about how you’re triggering the first message and I can help you smooth that out. @DomT

@Diana precious hi, I am not triggering the first message. It's set to "assistant-speaks-first", so it gets played automatically.

Please share your call id, so we can look int the issue "first message not triggering"

The message is properl recorded in vapi side. I assume it's a problem with telnyx. I will contact them.

sure, thank you for notifying.