Meta brings real-time AI avatars to the Muse app in the US
A photo, illustration, animal or object can become a live conversational character with synchronized voice and motion.
Meta is rolling out Muse Realtime Avatar in the Muse app in the United States, giving personal agents an expressive visual presence during live conversations.
A user can provide a portrait, full-body illustration, animal or everyday object as reference media. The system then generates synchronized speech, facial expressions, hand gestures and body motion while trying to preserve the character across conversational turns. Muse Realtime Voice and the avatar model share the same stream of speech tokens, which Meta says keeps voice, lip movement and expression aligned.
The company reports roughly 870 milliseconds between the end of a user’s turn and the first byte of the synchronized voice-and-video response. Output streams at 448×768 pixels and 25 frames per second. Meta says one GB200 can support about 12 concurrent video sessions after quantization and inference optimizations.
The rollout turns a research demo into a consumer feature for creating live fictional or personalized companions. Meta says generated video includes an invisible Video Seal watermark, and Muse is restricted to users aged 18 and older.
