Synthesia Voices page previews your script: test voices with Sume TTS

Synthesia's updates page for July 2026 describes a Voices page where you preview stock voices by language with your own script and download audio clips. If you work in Sume, the equivalent is to call the text-to-speech API with your script before committing to a video render. TTS costs $0.0475 per 1,000 characters, so a test of a 300 character script costs about 1.4 cents.

Why audition first

The voice is half of an avatar video. A voice that sounds wrong forces a re-render, and a 15 second avatar clip at standard quality is $2.76 against pennies for the audio test.

Language mismatch is a hard stop

The Sume TTS API notes that a voice and text language mismatch returns 409 tts_voice_language_mismatch. You can override with confirm_language_mismatch, but do it only when the accent is intended.

TTS input is capped at 20,000 characters per request, which is far above what a 60 second avatar script needs.

Then render

Once the audio is right, write the final script into the avatar request. Remember that duration is estimated from the script and must land between 4 and 60 seconds.

Keep the tested script in version control so the audition and the render use exactly the same words.