Multilingual AI avatar video, from one recording

Record once in any language. Then have the same person speak to every market, either in their cloned voice or with audio you've already dubbed.

Language per request

Set voice.language to an ISO 639-1 code and we voice the script in it.

Bring dubbed audio

Send finished speech in any language and the avatar speaks it as is.

Timings included

Word timings come back for captions and subtitles in every language.

Two ways to go multilingual

Text mode is the fastest: send the translated script with a language code, and we synthesize it in the avatar's cloned voice. Audio mode gives you full control: use your own dubbing pipeline or voice provider and send the finished track.

  • Text or script mode: 1.5 tokens per second, voiced in the avatar's cloned voice
  • Audio mode: 1 token per second, up to 10 minutes per request
  • Character timings in ElevenLabs format are accepted as is

Where teams use it

Training platforms updating courses for every region, dubbing and localization tools that need the original speaker on camera, and SaaS products onboarding customers in their own language.

Questions about multilingual video.

Which languages are supported?

Our voice engine speaks dozens of languages, chosen per request with an ISO 639-1 code. In audio mode any language works, because you send the speech.

Does the speaker need to record in every language?

No. One recording makes the avatar; the language comes from the script or the audio you send.

Can I keep using my own voice provider?

Yes. Audio mode is built for that: send your provider's output and optional timings, and pay 1 token per second.