Meta rolls out Muse voice and expressive avatars for real-time US chats

Meta pitches voice as a way to think out loud without polishing a prompt: sketch a trip, rehearse an interview, or dictate tasks while walking or cooking while Muse keeps working in the background[1]. Muse Realtime Voice is meant to answer immediately and stay interruptible, so the conversation can cut both ways[1].

As Muse speaks, its avatar uses expressions and gestures matched to the tone of the chat, and it reacts while you talk so it is clear the model is following along[1]. Presence is optional: voice only, avatar beside the transcript, or full screen for a face-to-face view[1]. Meta also offers hundreds of voices, from everyday styles to more theatrical ones, and says you can switch whenever you like[1].

The US Muse app rollout is the first surface. Meta says voice conversations with Muse are coming soon to AI Glasses and Muse Charm[1].

Text agents already handle planning and coaching; putting interruptible speech and a reactive face on the same loop is Meta's bid to make Muse feel like a companion you can talk to while your hands are busy, not only a chat box you open at a desk[1][2].