Build a lip-sync shot

Lip sync is a two-input workflow: a source video provides the visible speaker and dialogue audio provides the performance timing. The result is a new video, so the original stays available for comparison.

Choose footage that can synchronize well

Use a clear, visible face with an unobstructed mouth. Avoid rapid cuts, extreme profiles, very small faces, or motion blur when you can. Confirm that you have permission to use both the footage and the person's likeness.

Finish the audio first

Use clean dialogue with minimal noise, music, echo, or overlap. Review pronunciation, pauses, pace, and emotional fit before connecting it. Lip sync follows the supplied performance; it is not a substitute for directing the audio.

Connect the source video and dialogue audio to their named inputs. Select from compatible models shown in the node, review access and estimated credits, and generate.

Evaluate the new shot

Watch the mouth at normal speed, then inspect difficult consonants, pauses, and the beginning and end of speech. Check face stability, identity, background motion, and audio alignment. Compare against the source to find changes unrelated to the lips.

If the run disconnects, check its node status before retrying. The provider may still complete the work after the browser connection closes.

Use the approved lip-sync result as a source clip in an Assembly node when it belongs in a larger sequence.