Sync-3 launches studio-grade lip sync model
sync-3 is a 16B-parameter AI lip sync and visual dubbing model from Sync that aims at production workflows rather than toy demos. It generates full sequences in one pass, preserving performance detail across close-ups, occlusions, extreme angles, low light, and rapid dialogue, while supporting 95+ languages and 4K output. The product is available on sync.so, via API, and through an Adobe Premiere plugin, positioning it for creators and studios that need practical localization and finishing tools.
Strong launch if the claims hold up in real-world footage. The pitch is not “better lips,” it’s “better scene understanding,” which is the right framing for post-production buyers who care about consistency across an entire shot.
- –Full-shot generation is the key differentiator; stitching snippets is where most lip sync pipelines fall apart.
- –The workflow coverage is broad: web app, API, and Premiere integration lowers adoption friction for both studios and independent editors.
- –95+ languages and 4K output make this feel aimed at localization and finishing, not just social media remixing.
- –The hard test is whether it stays stable on occlusions, profile shots, and expressive acting without flattening emotion or identity.
- –If the quality matches the demo, this competes more with professional dubbing workflows than with casual avatar tools.
DISCOVERED
113d ago
2026-04-07
PUBLISHED
114d ago
2026-04-07
RELEVANCE
AUTHOR
[REDACTED]