Why separate the audio before lip-sync?

Isolating the voice track removes music and effects that can confuse the sync model, giving it a clean dialogue signal to align the mouth movement to.

How do I lip-sync the right person in a multi-character shot?

Drag a selection box around the speaking character and adjust it across the range of frames so it follows them as they move. The box tells the model which detected face to drive, so only that character is animated.

Can I lip-sync two characters in the same shot?

Yes — run lip-sync once per character. Select the first speaker over their lines, run it, then repeat with a new selection box around the second speaker. Each pass only animates the character inside its box.

Can I lip-sync a clip from any AI video tool?

Yes. Lip-sync operates on the video file itself, so a clip from any generator works as long as there’s a face and a dialogue track to sync to.

Any generator · Lip-sync prep & character selection

Prep AI video for lip-sync

Good lip-sync starts with a clean take, clean audio, and — when there’s more than one person on screen — the right character selected. Color Correct lets you trim the clip, pick the speaker, separate the voice track, and run the sync without bouncing between tools.

What you're seeing

Lip-sync results that drift because the input clip wasn’t trimmed cleanly
Music or effects bleeding into the dialogue and confusing the sync
Lip-sync animating the wrong person in a multi-character shot
Mouth movement that smears because the model locked onto the wrong face

Why it happens

Lip-sync models work best on a tightly-scoped clip with isolated dialogue, and they animate whatever face they detect. A full-length take with mixed audio — or a shot with several faces — gives the model extra, ambiguous signal, which shows up as soft, mistimed, or wrongly-targeted mouth movement.

How to fix it with Color Correct

Clip the exact range you want to sync so the model only sees the relevant take.
If the shot has more than one person, drag a selection box around the speaking character and adjust it across the range of frames so it tracks them as they move — only the character inside the box is animated.
Extract audio stems to separate the dialogue (“voices”) from music and effects (“sounds”).
Run Lip-Sync on the trimmed clip using the isolated voice track.
Recombine the synced video with your full mix afterward.

Free plan available — watermarked outputs, no card required.

Step-by-step in the docs

Frequently asked questions

Why separate the audio before lip-sync?: Isolating the voice track removes music and effects that can confuse the sync model, giving it a clean dialogue signal to align the mouth movement to.
How do I lip-sync the right person in a multi-character shot?: Drag a selection box around the speaking character and adjust it across the range of frames so it follows them as they move. The box tells the model which detected face to drive, so only that character is animated.
Can I lip-sync two characters in the same shot?: Yes — run lip-sync once per character. Select the first speaker over their lines, run it, then repeat with a new selection box around the second speaker. Each pass only animates the character inside its box.
Can I lip-sync a clip from any AI video tool?: Yes. Lip-sync operates on the video file itself, so a clip from any generator works as long as there’s a face and a dialogue track to sync to.

Related guides

Stop fighting color drift

Color Correct matches your AI video clips frame-to-frame so a sequence stays consistent.