Key Points
- Meta launched real-time voice conversations and an expressive avatar inside its Muse assistant app.
- The assistant can work on tasks while the user keeps talking, Meta said.
- Voice shifts the assistant off-screen, ahead of a promised rollout to Meta's glasses.
The latest:
Meta’s Muse assistant can now be spoken to in real time, with replies arriving instantly and the option to interrupt mid-answer, the company announced on November 5, 2026. Meta said the assistant can also carry out tasks while a conversation continues, and paired the voice mode with an animated avatar that reacts as it listens. The features were previewed at Connect 2026.
Details:
- How it starts: According to Meta’s newsroom post, a voice conversation begins by tapping the phone icon in the top right corner of the Muse app. Responses are delivered immediately rather than after a pause, and the company said users can cut the assistant off mid-sentence the way they would interrupt a person.
- Background tasks: Meta said Muse can handle requests while the conversation is still running, so instructions can be given while walking or with hands occupied. The company framed this as the core shift: the assistant is meant to be used without looking at a screen.
- The avatar: Alongside voice, Meta introduced an expressive avatar that reacts while listening and speaks with facial expressions and small gestures the company says are matched to the tone of the conversation. It is presented as a visual companion to the voice mode rather than a separate product.
- Display modes: Three viewing options were announced: voice only, the avatar displayed beside the chat thread, or the avatar in full screen. Meta did not describe any difference in capability between the three modes.
- Voice options: Meta said Muse offers hundreds of voices and speaking styles, citing a New York accent and a pirate voice as examples, and that users can switch between them at any time. Hundreds is the only figure in the announcement.
- The technology: The company attributed the features to two systems it calls Muse Realtime Voice and Muse Realtime Avatar, described in a linked research post. The newsroom announcement itself does not detail model size, latency figures or training data.
- Named use cases: Meta listed brainstorming, trip planning, practicing job interviews, rehearsing presentations and running through investment pitches as intended uses, a set weighted toward conversational rehearsal rather than search or information retrieval.
- Availability: The features are live in the Muse app on phone, web and Mac, rolling out in the United States first. Meta did not name additional countries or give a timeline for international availability.
- What’s coming: Meta said voice conversations are coming soon to its AI Glasses and to a device it calls Muse Charm. No release date was attached to either, and the announcement carries no quotes from named executives, appearing in the newsroom’s own voice.
Background:
Meta demonstrated real-time conversations with Muse at its Connect 2026 developer conference. The November announcement is the public release of what was shown there, rather than a new capability disclosed for the first time.
Between the lines:
The use cases Meta chose to list — interview practice, presentation rehearsal, pitch run-throughs — point at sustained back-and-forth sessions rather than one-shot queries, which is what interruptible, instant replies are built for. The background-task claim and the promised move to AI Glasses and Muse Charm suggest the phone app is a staging ground: the behavior being tested now is hands-free use, before the hardware that depends on it ships.
What’s next
Watch for a release date and pricing for Muse Charm, the arrival of voice on Meta’s AI Glasses, and any expansion of availability beyond the United States.
