Skip to main content
Percify’s lip-sync turns a still photo and an audio track into a talking video: the face in the photo is animated and its mouth follows the audio. The audio can be a recording you make in the browser, an audio file you upload, or a clip saved in My Voices. You pay per second of audio, from 2 credits a second on Fast to 6 on High at 720p.

​
What you need

  • A computer, with the desktop web app at app.percify.io.
  • A photo with a clear, front-facing face.
  • The audio: a recording of 5 to 60 seconds, an audio file under 10 MB, or a saved clip.

​
How to lip-sync a photo

1

Open lip-sync

On the Explore page, click Lip Sync Video under Utility, or open Clone Yourself in the sidebar. Both start by asking for the photo.
2

Add the photo

Click Upload Image or Use Camera, check the preview and click Continue. To use a ready-made face instead, start from the avatar library.
3

Add the audio

On Create Talking Avatar, choose the source:
  • Record: tap the microphone, speak for 5 to 60 seconds and stop.
  • Upload Audio (on the Record tab): pick an audio file under 10 MB.
  • My Voices: pick a recording or a generated clip you saved before, for example speech in your cloned voice.
4

Pick mode and quality, then generate

Choose Fast or High, and 480p or 720p (720p needs High). When Estimated cost shows the credits, click Generate video. When it is ready, click Download.

​
How much does it cost?

The cost is the per-second rate times the length of the audio. The server measures the audio file itself and rounds up to a whole second, so the charge follows the real length, not the timer. The estimate appears as soon as the audio length is known (“Estimating cost…” means it is still loading). If your balance is too low the button reads Not Enough Credits and nothing is charged. See how credits work, and refunds if a video fails.

​
Audio requirements

​
Lip-sync in the Playground and the API

The Playground lists the same lip-sync models as InfiniteTalk Fast (2 credits a second, minimum 4) and InfiniteTalk (4 credits a second at 480p, 6 at 720p, minimum 8). Each takes an image and an audio file. Use them to run lip-sync next to other models, or to copy an API call from the model page. API access requires the Scale or Ultra plan; see the API introduction.

​
Tips

  • Trim silence from the start and end of an uploaded file. You pay for every second of it.
  • Test a short clip on Fast before a long one on High.
  • If the mouth drifts or the face does not move, try a sharper, front-facing photo and cleaner audio. See lip-sync errors.
Not in this generator: it animates a still photo. To change the language of speech in an existing video, use the video translation models described in Dub a video.
The charge is based on the length of the saved audio file, measured on the server. A browser recording can run slightly longer than the timer showed.
They are listed on the History tab of Clone Yourself, and you can download each one.

Make a talking avatar from a photo

Start from a ready-made face in the avatar library.

Create your AI twin

Lip-sync a photo of yourself to your own voice.

Lip-sync errors

Fixes when the result looks out of sync.

Clone your voice

Make audio in your voice from typed text.
Last modified on September 16, 2026