ex-omni-avatar syncs voice and talking-head output from one model

victormustar · x · 2026-07-25

HuggingApps says ex-omni-avatar points toward a truly omni-modal future: one model takes user input and produces both the spoken answer and the talking head output in sync.

The demo setup is not yet a realistic face rig, but the post says that part can be swapped in for real applications.

Original post →

More from Multimodal

Multimodal channel →