Dubbing a clip used to mean lining up mouth shapes by hand. The free AI lip sync tool on ToolXoX does that job from two uploads: a video of a face, and the audio you want the mouth to follow. You generate a new clip in the browser, without an After Effects project and without an account. The usual download does not pick up a watermark from the page.
The tool is built for speech and singing when a face stays clearly in view. It works with real people and with many AI avatars. Animals and some cartoons are a weaker match, because the model is estimating human mouth shapes from the sound.
What you can upload
The first slot is the face video: MP4, MOV, or WEBM, from 10 seconds to 1.5 minutes, up to 500 MB. The second slot is the audio you want to sync: MP3, WAV, or M4A, up to 50 MB. The audio can be in any language. Trim that track to exactly the line the person should say or sing. Extra silence, a second speaker, or a music bed under the voice all make the mouth harder to follow.
A steady, front-facing shot in good light is the best source. If hands, hair, or a microphone hide the mouth, or the head turns away for a long stretch, the sync will slip.
How to lip sync a clip, step by step
- Prepare the video. Cut it so the face is visible for the whole take, and keep the length between 10 seconds and 1.5 minutes.
- Prepare the audio. Export a clean, single-voice track as MP3, WAV, or M4A, and trim it to the words or the phrase you need. Stay under 50 MB.
- Upload both files in the correct slots — face video first, audio second — and check that you have not swapped them.
- Complete the security check and click Generate once. The job is queued and processed frame by frame, so a longer clip takes more time. Leave the tab open.
- Watch the result all the way through before you download it. Check the start of each sentence, the ends of words, and any stretch where the head moves.
What a clean sync needs
Lip-sync models estimate mouth shapes from the audio. Fast speech, several voices at once, or a song with heavy backing vocals will throw the estimate off. A clear voice recording, one speaker, and a face that stays toward the camera give the most believable result.
Match the performance to the new line when you can. A calm talking-head is easier to dub than a shout or a wide laugh that does not fit the new words. If you are localising a demo, keep the new script close in length to the original take so the mouth is not left hanging at the end of the shot.
Related tools for talking faces
If you only have a still photo, not a video, you do not need lip sync. The talking photo tool on CanvaPro animates a portrait from text or audio, and there is a full walkthrough on how to make a talking photo. If you want the whole picture to move, not just the mouth, image to video is the closer path, and you can follow a guide to turning an image into a short video.
A finished talking-head that needs a different face, rather than a new voice, belongs on a video face swap, not on this page.
The time window is strict. A clip shorter than 10 seconds does not meet the limit, and a take over 1.5 minutes needs a trim before you upload. If the new audio is longer than the video, cut the line or extend the shot so the mouth has a frame to finish on. Multi-language audio is supported, so a dub in another language is a normal use, as long as the track is still a single, clear voice.
If the mouth does not match the sound
Re-trim the audio so it starts at the first word, and use a take where the mouth is not hidden. If the clip is barely over 10 seconds, a slightly longer video can help; if it is close to 1.5 minutes, cut it down and run the important section on its own. An upload error is often a format or size issue — stay with MP4, MOV, or WEBM for video, and MP3, WAV, or M4A for audio.
Watch the preview before you publish. A small slip at the end of a line is easier to hide with a cut in your editor than with a second, rushed generate.
When the face and the line are ready, open the lip sync generator, upload both files, and download the clip only after you have watched the mouth against the new audio.