Meta Unveils Muse Realtime Avatar: Speech-Driven Real-Time Avatars
alex_conneau · x · 2026-09-25
Meta AI introduces Muse Realtime Avatar, extending Muse Realtime Voice into expressive real-time interactive avatars.
- Capabilities: Conditioned on reference media, any character — photographic portraits, full-body illustrations, animals, objects — comes alive with facial, hand, and full-body motion, staying coherent across conversational turns
- Architecture: Voice and Avatar share a single streaming speech token (VQ) stream for synchronized speech, lip motion, and expression; the avatar is an audio-driven Diffusion Transformer generating short causal chunks with a rolling window of recent latents, bounding computation so generation can run as long as the conversation
- Led by Amaury Sabran over nearly two years (started at Waveforms AI), solving lip-sync, distillation, long-rollout consistency, efficiency, hand gestures, and any-style support; former Meta voice lead Alexandre Conneau publicly praised the work
Related event: Meta Unveils Muse Realtime Avatar: Sub-second Realtime Digital Humans(18 posts)→
More from Multimodal
- Ditto launches swap-anything video tool to replace any video's main character — yoobinray · 2026-11-12
- Fable 5.2 impresses with pixel art gaussian splats, posts go viral — Flomerboy · 2026-11-12
- Seedream 5.5 nails a consistent character sticker set at just $0.02 per image — azed_ai · 2026-11-12
- Ant Group's open-source Ming-3.0-tts-0.5B tops Open TTS Leaderboard with 0.88 WER — NielsRogge · 2026-11-12
- Viggle Remake recreated in ComfyUI: swap any character in your videos with MiniMax H3 — TheLocalLab · 2026-11-12
- Pika Integrates Seedream 5.5: Split Images Into 100+ Layers for Precise Multi-Turn Edits — aziz4ai · 2026-11-12