Lip-sync from a photo: an AI talking head
Lip-sync needs two things: a photo of a person and an audio track, for example a voiceover. The model syncs the lips and facial motion to the voice. You can also upload a finished clip with a person, and the lips will follow the new speech. The sharper the face in the photo, the better the mouth motion.
Updated:
Step by step
Portrait
Face forward or three-quarter, no glasses and no hand near the mouth. The model then sees the lips fully.
Audio
A voiceover from the Voiceover section or your own recording. Hiss and music throw off the sync, so clean speech works best.
Run
Pick the input: photo and audio, or video and audio. The total shows on the button before you start.
Where it is in ClickBait
Frequently asked questions
Can I use a clip instead of a photo?
Yes, there is a video and audio input. The lips follow the new speech.
Why are the lips a little late?
Usually background music or long pauses cause it. Remove the noise and run it again.
Do I need a separate voiceover?
You need a track with speech. Make it in the Voiceover section or record it yourself.
Can I lip-sync several lines?
Yes, run each line separately and join them in editing. That makes it easy to fix one line without redoing the rest.
Does a photo with a smile or an open mouth work?
A neutral face works more reliably. An open mouth in the photo can throw off the lip motion, so a closed-mouth photo is better.