Meta details Muse Realtime Voice: VQ tokens drive parallel audio and video decoders
alex_conneau · x · 2026-09-24
Muse Realtime Voice continuously generates VQ tokens: a speech decoder reconstructs audio while an audio2video decoder produces video chunks for realtime avatars. The author calls diffusion-based videogen for avatars "brute force" but argues the same heavily optimized architecture can extend to richer voice-driven interactive worlds — realtime video adds expression, embodiment and context on top of voice.
Related event: Meta Unveils Muse Realtime Avatar: Sub-second Realtime Digital Humans(18 posts)→
More from Models
- antirez backs low-bit DeepSeek 4.1 Flash over Qwen 27B dense on Apple M5 Macs — antirez · 2026-11-12
- Gemini re-renders images as text when vision is removed, dropping only 6% WSR — zacharynado · 2026-11-12
- Bare-harness coding benchmark: Fable 5.2 best overall, Gemini 4 Carbon best value — Semantic_meaning · 2026-11-12
- Google's Physical-Engineering's Last Exam: all 14 frontier models score below 50% — dl_weekly · 2026-11-12
- Reddit's 2026 LLM hall of fame: Opus 5.5 still unbeaten after 9 rival releases — G0dZylla · 2026-11-12
- Commit leak reveals unreleased Gemini 4 variants: Argon, Hydrogen and Helix (unconfirmed) — BlackHC · 2026-11-11