title

VAPI TTS models can support SSML if you use a custom TTS integration and your underlying TTS provider supports it. You can preprocess SSML tags in your custom TTS server as shown below:

javascript
// Handle SSML tags if your TTS supports them
if (text.includes('<speak>')) {
  return parseSSML(text);
}

To enable SSML, ensure your TTS backend can interpret SSML input and integrate it with VAPI's custom TTS webhook. For built-in VAPI voices, native SSML support is not explicitly documented. Source: - [Custom TTS integration documentation](https://docs.vapi.ai/customization/custom-voices/custom-tts)

The VAPI tts models and voices sound excellent. Unfortunately, they handle recitation of numbers and characters poorly. I want to add pauses and recitation speed changes with ssml. This is specifically for the vapi provider

Currently, VAPI's built-in TTS voices do not natively support SSML for adding pauses or changing recitation speed. If you require SSML features, the recommended approach is to use a custom TTS integration where your backend can process SSML tags and return the desired audio. For now, with the VAPI provider, there is no documented way to control number/character recitation using SSML. See the [VAPI Voices documentation](https://docs.vapi.ai/providers/voice/vapi-voices) for more details. Source: - [VAPI Voices documentation](https://docs.vapi.ai/providers/voice/vapi-voices)

Message marked as helpful by @Lee Harrold! 🎉

solved