I’m currently working with a Vapi assistant that functions well overall, but we’ve been experiencing issues when using third-party voices (such as ElevenLabs and PlayHT). The assistant works perfectly with the built-in Vapi voices, but the third-party options aren’t performing as expected. These voices usually have breaks in the voice while AI is speaking. We’re also using Make.com for automation in our setup. Could you please assist us in resolving this issue?

Ensure third-party APIs are correctly set up, switch to more stable voice models, use SSML to smooth speech, optimize your Make.com workflows to reduce delays, and consider using built-in Vapi voices for live calls while reserving third-party voices for non-live messages.