Lip sync and avatars models
Performance transfer onto a face that did not say it.
- Catalogued here
- 8 models
- Open weights
- 1
- With an API
- 6
What to know about lip sync and avatars
Performance transfer onto a face that did not say it. Technically the most mature category on this list and legally the most exposed, because the subject is somebody's likeness. Consent requirements are the first thing to check on every page here.
Duration and resolution, side by side
Only the models whose vendor publishes a figure appear here. "Not established" means no page we could retrieve stated one.
| Model | Duration | Resolution | Weights | Commercial use |
|---|---|---|---|---|
| HeyGen Avatar IV (and Avatar V) | 3600 s | 3840x2160 | Not released | Permitted subject to conditions |
| lipsync-2 and lipsync-2-pro | 1800 s | 512x512 | Not released | Not established |
| sync-3 | 1800 s | 3840x2160 | Not released | Not established |
| LatentSync | not established | 512x512 | Downloadable | Permitted subject to conditions |
| Synthesia Express-2 | not established | 1920x1080 | Not released | Permitted subject to conditions |
All 8
-
Flawless DeepEditor and TrueSync
Neural performance editing tools for film and episodic post. DeepEditor applies new dialogue to existing shots and refines performances, covering in-vision ADR, censorship editing (replacing
-
Hedra character / avatar platform
Character and avatar video platform that has grown into a unified inference API putting multiple leading image, video and audio models behind one API key, one endpoint and one bill. The deve
-
HeyGen Avatar IV (and Avatar V)
Avatar IV is HeyGen's current default avatar engine, driving a photo, consented digital twin or studio avatar from either a typed script (text to speech) or an uploaded audio file, analysing
-
LatentSync
Open-weights end-to-end lipsync method built on audio-conditioned latent diffusion, generating mouth movement directly from audio without an intermediate motion representation. Version 1.6 w
-
Mirage / Captions
Mirage is a family of end-to-end audio and video foundation models that generate performances, voiceover, captions, graphics and music score, with the output kept as editable layers rather t
-
Synthesia Express-2
Combined video and voice engine that generates full-body avatars which gesture, not just talk. Express-Voice clones a voice in context without fine-tuning, preserving accent and expressivene
-
lipsync-2 and lipsync-2-pro
lipsync-2 is Sync's general-purpose lipsync model, preserving each speaker's individual speaking style at 512x512 resolution. lipsync-2-pro adds diffusion-based super resolution for better d
-
sync-3
Sync's flagship lipsync model and the current default for all users. Processes an entire shot holistically rather than in snippets, which lets it handle occlusions, extreme angles and close-