Avatar & Voice
Talking Portrait
Upload one photo. It speaks any language.
Give it a portrait and a script and it returns an 8-second clip of that face speaking your words with real mouth-sync — in whatever language you write them in. Built on veo-3.1-fast, which has no language setting to limit you: it renders Azerbaijani speech natively, lips and all, instead of dubbing it on afterwards.
Frequently asked
Does it really speak Azerbaijani, or is it dubbed?
Really. The model renders the speech and the mouth together, so the lips match the Azerbaijani words. Nothing is dubbed on afterwards.
Which languages can it speak?
Dozens — pick one from the Spoken language list. The model has no fixed language menu of its own: it speaks the language of the script you write. The interface ships in three languages; that has nothing to do with what the clip says.
Can I use a photo of someone else?
Only with their permission — you confirm that before the first render. Public figures and photos you do not own are not allowed.
Why did my render come back refused?
Scripts that put words in a named real person’s mouth get refused by the model. Keep the script to your own message and use your own photo.
Related tools
- Ad in 60 — A product photo in, a finished ad out.
- Podcast Clip — An audio file becomes a captioned vertical clip.
- Talking Avatar — Photo + script = video.