Hi team, I have an reception agent that ensures it captures details correctly by repeating back the user's input. Eg "Just checking, your name is J O H N S M I T H and your number is XXXXXXXXXX Is that correct?" I'm getting a weird bug where at the end of the sentence, when it gets to "is that correct" the speech generation will go haywire and play artifacts like AHHHH and speek gibberish before finishing the sentence. Call ID: - 0199b2e2-bc84-7774-82a9-4d204fbf8a3c

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

Try sanitizing your TTS input text to remove any problematic characters or formatting that might cause synthesis artifacts. Here's a sample function you can use before sending text to your TTS system:

js
function preprocessText(text) {
  return text
    .replace(/[^a-zA-Z0-9\s.,!?-]/g, '') // Remove special characters
    .replace(/\s+/g, ' ') // Normalize whitespace
    .trim();
}

Test your TTS endpoint with this cleaned text to see if the issue persists. For more, see the [custom TTS troubleshooting guide](https://docs.vapi.ai/customization/custom-voices/custom-tts#troubleshooting). Source: - [Custom TTS integration](https://docs.vapi.ai/customization/custom-voices/custom-tts)

I'm currently working on the theory that repeating the full name + number is too long for the agent to handle...

We recommend lowering the assistant temperature to 0.2 then if that doesn't work, switch to a different voice provider or LLM model such as GPT-4o