Upload a clip and the speech you want. AI lip sync redraws the mouth so each word lines up, for dubs, new lines and AI characters.
You get an MP4 with the new speech and matching lips. Most short clips take 2–3 minutes.
One of our own tests, start to finish. Switch between the silent clip and the lip-synced result, with sound on.
The line he says: “Hi there. A minute ago this clip was silent. Now every word I say matches my lips.”

A photo of a person who does not exist: face to camera, mouth closed, even light.

Eight seconds of blinking and small head moves, generated from the photo with Veo 3.1 Lite.

The AI lip sync redrew his mouth for every word. The rest of the frame stayed as it was.
AI lip sync is a tool that changes the mouth movements in a video so they match a new audio track. You give it footage of a person and the speech you want, and it redraws the lips, jaw and teeth frame by frame until the words and the mouth line up.
It saves a reshoot. A dubbed ad, a corrected product name or an AI character who needs a voice used to mean filming again or animating by hand. AI lip sync does it from the clip you already have.
The model listens to the audio and splits it into phonemes, the sounds of speech. Each sound maps to a mouth shape, called a viseme: lips closed for M, wide for A, round for O. The model then repaints the lower face in every frame to hit those shapes on time, blending them into the skin and lighting around it.
Framing, background, lighting, head movement and the rest of the face stay as they were. Only the mouth area is redrawn.
Lips, teeth and jaw follow the new speech, and the new audio replaces the clip's original sound.
The whole AI lip sync runs in your browser. No editing software and no plugin.
Drop an MP4 or MOV of 2 to 60 seconds. One person facing the camera, with the mouth visible, gives the best result.
Upload MP3, WAV, M4A or AAC audio up to 10 MB. Record it on your phone, export it from a voice tool or use a dub.
Tick the consent box and press Sync. Most short clips are ready in two to three minutes as an MP4.
AI lip sync can only redraw a mouth it can find. These five things decide whether it finds one.
Front-facing or slightly turned works best. A full side profile hides half the lips.
The model needs both corners of the mouth to place each shape.
Use at least 720p for close-ups. Very small or blurry faces give soft, smeared lips.
More pixels around the mouth means more detail to redraw.
Keep one clear face in the shot. With a crowd, the sync may land on the wrong person.
The tool drives a single face per clip.
Hands, microphones, masks and thick beards over the lips get in the way.
Anything covering the mouth is redrawn badly or not at all.
Fast head turns and flickering light make the mouth hard to track from frame to frame.
Motion blur hides the shape the model is trying to change.
The result keeps your video's length, and you pay for the video's seconds. Here is what happens when the two files don't match.
| Your files | What you get | Tip |
|---|---|---|
| Audio as long as the video | Lips move for the whole clip | The ideal case for dubbing |
| Audio shorter than the video | The speaker stops talking when the audio ends | Trim the video to the audio first and pay for fewer seconds |
| Audio longer than the video | The audio is cut where the video ends | Speed up the line or use a longer clip |
The tool shows both lengths before you sync, so a mismatch is never a surprise.
There are three ways to get a video of someone saying your words. The right one depends on what you start with.
Price on Imgveo
40 credits for a 5 s clip
Price on Imgveo
From 30 credits for an 8 s clip with sound
Prices are read from our live price list. The talking photo row is the exact route we used for the example above.
AI lip sync is the last step of a dub. The same footage can speak Spanish, Japanese or German without a new shoot.
Translate the line and read it aloud. Keep it close to the original length.
Record a native speaker or export the line from a voice tool as MP3 or WAV.
Trim silence at the start and end so the speech fills the clip.
Upload the original video with the new audio. Repeat for each language.
Some languages need more syllables for the same idea. If a German line runs long, shorten the wording rather than speeding up the voice, which sounds rushed. AI lip sync follows whatever pace the audio has.
Anywhere a face on screen needs to say something it didn't say on the day.
Change a price, a product name or a call to action without booking the presenter again.
Keep a clean take of the presenter looking at the lens for future edits.
Ship one video in several languages with mouths that match each one.
Run the most important market first and check the result before the rest.
Give a generated character a voice and a line of dialogue that fits your story.
Generate the character facing the camera with a closed mouth, as in our example.
Swap one wrong word in a talk, a course or a social clip without reshooting.
Cut out just the sentence, sync it, and splice it back into your edit.
We tested the limits on our own clips. Here is what to expect.

If the face is turned away, tiny or blurred by motion, the sync stops and your credits come back.
One face is synced per clip. For a conversation, cut each speaker into their own clip.
AI lip sync matches the mouth, not the delivery. A flat recording still sounds flat.
What the AI lip sync accepts, what it returns and what it costs.
Video: MP4 or MOV, 2 to 60 seconds, up to 100 MB, at least 360p. Audio: MP3, WAV, M4A or AAC up to 10 MB.
An MP4 at the video's own length, with the new speech as its soundtrack and the mouth matched to it.
8 credits per second of video, rounded to whole seconds. Paid plans only. Failed runs are refunded.
| Video length | Credits |
|---|---|
| 5 s | 40 |
| 10 s | 80 |
| 30 s | 240 |
| 60 s | 480 |
AI lip sync can put words in someone's mouth, so we ask for permission first. Before every run you confirm that you have the right to use the face and the voice.
Go deeper: guides from our blog
Open Imgveo Studio to run this and every other AI video and image model from a single workspace, with one credit balance.
Make the clip, then give it a voice.
Turn a portrait into a short clip you can lip sync.
Animate a photoTake your lip-synced clip to 1080p, 2K or 4K.
Upscale a videoGenerate new scenes with dialogue and sound built in.
Try Veo 3.1Upload a portrait and a voice track, and the AI talking photo maker does the rest in a few minutes.
Animate a photoUpload a clip and the speech you want, and the AI lip sync does the rest in a few minutes.