Two distinct inputs
Lip sync uses a video clip and an audio track; a still image is not the final video input for the documented workflow.
VIDEO AND AUDIO WORKFLOW
AI lip sync for video and speech
Work from a video clip and clean audio track, then watch the generated result from start to finish. The three-frame image is an illustrative storyboard, not a playable or verified lip-sync output.
How to use AI Lip Sync
- 1
Prepare the two inputs
Choose a face video you have permission to use and a clean speech recording. If you only have a portrait, make a base video first.
- 2
Connect video and audio
On the Astorie canvas, add the lip-sync tool and connect the video and audio to their matching inputs.
- 3
Review the entire clip
Watch with sound. Check consonants, pauses, the final word, face consistency, and any visible artifacts before using the result.
What to focus on
Timing needs review
A completed render can still miss mouth shapes or pauses. Watch the result at normal speed with the original audio.
Use permitted likenesses
Only use people, voices, and footage for which you have the necessary consent or rights.
Inputs, output and review
| Input | A face video plus a separate speech track in the documented canvas workflow. | |
|---|---|---|
| Output | A video to review with sound; the preview image on this page is only an illustration. | |
| Quality check | Inspect lip timing, identity, blinking, pauses, and the end of the clip. |
AI Lip Sync FAQ
Can I lip sync a still portrait?
The documented Kling LipSync A2V workflow requires video and audio. If you begin with a portrait, create a short base video before the lip-sync step.
Does lip sync guarantee natural dialogue?
No. Review the entire render for mouth timing, expression, and visual changes before publishing it.
Where do I run the lip-sync step?
Open the Astorie canvas, add the lip-sync tool, and connect your video and audio there. This page does not present a two-input lip-sync widget.
Build a lip-sync workflow on the canvas
Connect permitted video and audio, generate a short test, and review it with sound.