Ijharul Islam
To lip sync a video to new audio with AI, upload a clip where the speaker's face is clearly visible, upload the new audio track, and run a lip sync model. It redraws the mouth area frame by frame so the lips match the new words, while the rest of the video stays the same.
This is how people fix a fluffed line without reshooting, swap in a translated voiceover, or update a price in a product video. This guide covers how to prepare the video and audio, which models to use, the costs in QuickVeed, free options, and the consent rules.
The short answer
Use a clip with the face front-on, well lit and unblocked, and a clean audio track about the same length. Upload both to an AI lip sync tool, pick a model, and generate. Check the mouth on close-ups and the first and last seconds. Only lip sync people who agreed to it, and never make anyone appear to say something they didn't approve.
Key takeaways
- The video needs a clear, mostly front-facing view of the mouth; side profiles and hands over the face fail.
- Clean audio without music gives the most accurate mouth shapes.
- Match the audio length to the video, or trim one to fit, before generating.
- Cheaper models are fine for small faces in the frame; pay more for close-ups.
- Get the speaker's consent, and label realistic results where platforms ask.
Disclosure: we make QuickVeed and show how to do this in it, along with other ways. Details about other products come from their public pages, checked in October 2026 and linked at the end.
When AI lip sync is useful
- Fixing a line: the speaker said the wrong date or price. Record the right line, sync it, no reshoot.
- Translation: pair a dubbed voice with matching mouth movement so viewers aren't distracted by mismatched lips.
- Personalised video: the same presenter clip with a different name or offer in the audio.
- Animated avatars: making a generated or stock presenter say a new script.
Lip sync changes only the mouth area. It doesn't change expressions, gestures or head movement, so the new audio should have a similar tone and energy to the original performance.
Preparing the video and audio
| Video checklist | Audio checklist |
|---|---|
| Face visible and mostly front-facing | One speaker, no background music |
| Mouth not covered by hands, a mic or hair | No echo; record close to the mic |
| Steady, sharp footage at the final resolution | Similar pace to the original speech |
| Clip trimmed to the part you need | Same length as the clip, or trimmed to fit |
| Speaker's mouth closed at the start and end, if possible | Exported as WAV or high-quality MP3 |
Free browser tools help with the prep: extract audio from video pulls the original track so you can compare timing, mute video removes the old audio, and the video duration calculator helps you match lengths.
If you need a new voice track and can't record it yourself, generate it with text to speech, or clone the speaker's voice with their consent. See how to clone your own voice with AI.
Step by step
- Cut the video down to the section you want to change, plus half a second either side.
- Record or generate the new audio and trim it to the same length.
- Upload both to a lip sync tool and choose a model.
- Generate, then watch at full size. Check close-ups, teeth, and the moments of fast speech.
- Put the synced section back into the full video in your editor and check the cut points.
- Add captions so the text matches the new words.
Common problems and fixes
| Problem | Likely cause | Fix |
|---|---|---|
| Mouth looks blurry | Low-resolution source or cheaper model on a close-up | Use a sharper clip or a higher-end model |
| Lips lag behind the words | Silence or noise at the start of the audio | Trim the audio so speech starts where the clip starts |
| Mouth moves during silence | Background noise or music in the track | Use a clean voice-only track |
| Face flickers | Fast head movement or cuts inside the clip | Sync each shot separately, between cuts |
| New words feel rushed | Audio longer than the clip | Rewrite the line shorter, or extend the shot |
Lip sync for translated videos
Lip sync is most often used with translation. A dubbed voice over the original mouth movements looks off, especially in close-ups, and viewers notice. Syncing the mouth to the new language fixes that. The catch is length: a sentence in Spanish or German is often longer than the same sentence in English, so the new audio may not fit the original clip.
- Dub or record the translated audio first, keeping each line close to the original's timing.
- Pick the shots where the face is large and clearly visible. Wide shots and shots where the speaker is turned away rarely need lip sync.
- Cut those shots out, sync each one to its section of the new audio, and put them back.
- Leave the rest of the video with the dubbed audio alone. This keeps cost down and avoids touching shots that already look fine.
| Shot type | Lip sync it? |
|---|---|
| Close-up talking to camera | Yes |
| Medium shot, face clearly visible | Usually |
| Wide shot or small face | Rarely needed |
| Speaker turned away or off screen | No |
| Screen recordings and B-roll | No |
For the translation and voice side, see how to dub a video into another language.
Lip sync in QuickVeed
QuickVeed's AI avatars page includes a Lip Sync mode that syncs an existing video to new audio, up to 30 seconds per video. You choose from four models:
| Model | Credits per 10 s | 30 s clip | Good for |
|---|---|---|---|
| LatentSync | 37 | 111 | Tests, small or medium faces in frame |
| PixVerse Lipsync | 100 | 300 | General use |
| Lipsync 2 Pro | 209 | 627 | Close-ups where detail matters |
| Veed Fabric 1.0 | 375 | 1,125 | Highest-cost option; try it when others fall short |
- Open AI avatars in the dashboard and choose Lip Sync.
- Upload the video and the new audio.
- Pick a model and generate. The result is saved in your asset library.
Start with LatentSync on a short test. If the mouth looks soft or wrong on a close-up, move up a model. For a whole video in another language, AI dubbing handles transcription, translation and the cloned voice in one place; you can then lip sync the key close-ups.
Free options, limits and consent
LatentSync, the cheapest model above, is also open source: ByteDance released it under the Apache 2.0 licence, and you can run it on your own computer with a capable graphics card. Online avatar platforms such as HeyGen also offer lip sync as part of their tools.
Limits to expect with any AI lip sync:
- Side profiles, beards covering the lips, and very fast head turns reduce accuracy.
- Teeth and tongue can look blurred on large close-ups with cheaper models.
- If the new audio is much faster or slower than the original, the rest of the face won't match the energy.
Consent comes first. Lip sync can put words in anyone's mouth. Only use it on yourself, actors or staff who have agreed to it, or stock and AI presenters whose terms allow it. YouTube asks creators to disclose content that makes a real person appear to say something they didn't, and TikTok requires labels on realistic AI-generated video.
Frequently asked questions
Can AI change what someone says in a video?
Yes. You provide new audio and a lip sync model redraws the mouth to match it. Use it only with the speaker's permission.
What is the best AI lip sync model?
It depends on the shot. Cheaper models like LatentSync work for small faces; higher-end ones like Lipsync 2 Pro or Veed Fabric 1.0 are better for close-ups. Test on your clip.
Is there a free AI lip sync tool?
LatentSync is open source and free to run on your own hardware. Online tools usually charge credits.
How long can a lip sync video be?
It depends on the tool. QuickVeed syncs up to 30 seconds per video; split longer videos into sections.
Why does my lip sync look blurry?
The mouth area is regenerated, and cheaper models lose detail on close-ups. Use a higher-quality model, a sharper source video, or a wider shot.
Try it with your own video
Try a 10-second lip sync on a clip of yourself. Describe the video or paste your script, pick a voice, and judge the first draft for yourself. QuickVeed plans start at $18/month, and new accounts get a 14-day trial with 75 credits, no card needed. Sign up or see pricing.
Sources
Competitor details in this post were checked against these public pages in October 2026. Products and prices change, so confirm with each vendor before you decide.
