Lip sync on a moving mascot character ?

I’m looking for an AI solution capable of doing lip sync on a non-realistic mascot/cartoon character, using:

* either an existing video,
* or a starting image + prompt,
* and a .wav song/audio track to synchronize.

My goal:
→ obtain a mascot character in motion (walking/light dancing) WITH proper lip sync to the song.

Constraints:

* online solution only (no local setup)
* non-human / cartoon / mascot character
* ideally compatible with video-to-video or image + audio workflows

What I’ve already tested:

* Kling Lipsync V1 → does not correctly detect my character
* Sync.so → does not work well with mascot/cartoon characters
* Dreamina → works a bit, but the results remain limited
* Hedra → good lip sync, but difficult to add real body movement afterward

So I’m looking for a tool that properly handles:

* mascots / cartoon characters
* singing lip sync
* body movement
* character consistency

Even a “semi-working” or beta solution would interest me.

Thanks

Saturday bump !

Sunday bump !

i have a similar project and looking for the same thing as you, bookmarking this for future replies

cybersexuals said:

i have a similar project and looking for the same thing as you, bookmarking this for future replies

Click to expand...

I got a good result with jxp.com and the coupling with Infinite Talk V2V. This is precisely what might interest you. Unlike Kling or Hedra, which often focus on the face, Infinite Talk synchronizes the lips, head, body, and expressions. It's V2V (Video-to-Video): you don't start from a simple photo, but from an existing video (for example, of your AI character that is already moving). The tool will 're-animate' its body, making it adopt new gestures and movements synchronized with the new audio track. It is designed to avoid inconsistencies over long durations. It's perfect for songs or slightly longer dialogues.

no all-in-one yet that does mascot + motion + lipsync clean. best bet is stack tools. animate character in kling/runway, export, then pass frames into lipsync model that allows custom face mapping. messy but works