OPEN RESEARCH. REAL-WORLD LIP SYNC.
Research Models
Production Speed
We turn open-source lip-sync research into lean, production-ready runtimes. Optimized GPU execution, shared preparation and on-demand processing keep compute costs down.
From $0.10 / output minute
ONE PERFORMANCE. THREE PERSPECTIVES.
See The Difference
MuseTalk 1.5
Rebuilds the mouth region in latent space in one pass. A practical choice for fast iterations and high-volume dubbing.
One generation pass means less GPU work—and our lowest price.
LatentSync 1.6
Refines the face over multiple diffusion steps, using the audio and surrounding frames to guide lip movement and visual consistency.
Iterative refinement uses more compute than MuseTalk, with more room for fine detail.
HighSync
Pairs reference and denoising networks to generate a 512px face region while keeping appearance and movement tied to the source.
Two cooperating networks and repeated refinement put it in our highest compute tier.
LipForcing
Few-step autoregressive diffusion designed for real-time lip sync. Want it on Lips? Register your interest—if there’s enough demand, we’ll consider adding it.
Three approaches. One workflow. Prices per output minute.
Try Your Own Video →YOUR VIDEO + AUDIO. WE HANDLE THE REST.
Send It
We’ll Let You Know
01
Send Your Video And Audio
Upload in Studio, or send your media links to the API with the engine you want and a webhook URL.
02
Leave The Processing To Us
Lips handles the media, runs your selected engines and keeps track of the job. No preparation workflow to build.
03
Get Notified When It’s Ready
We call your webhook when the job finishes. Fetch the result and download your lip-synced video.
curl -X POST "$LIPS_API/v1/renders" \
-H "Authorization: Bearer $LIPS_KEY" \
-H "Idempotency-Key: render-001" \
-H "Content-Type: application/json" \
-d '{"video":{"url":"https://example.com/video.mp4"},"audio":{"url":"https://example.com/audio.wav"},"engines":["musetalk"],"sync_mode":"loop","webhook_url":"https://your-app.example/lips/webhook"}'