Meta puts expressive avatars and realtime voice in Muse

Staring at a blank prompt box is often where good ideas go to die. Meta wants you to skip the typing entirely.

Meta began rolling out realtime voice conversations and expressive avatars across the Muse app on mobile, web, and Mac. First previewed at Connect 2026, users can now tap the phone icon in the top right corner to instantly start talking with their AI.

Natural banter with mid-sentence interruptions

The update is powered by two distinct systems: Muse Realtime Voice and Muse Realtime Avatar.

Instead of the rigid turn-taking typical of earlier voice bots, Muse now lets you interrupt whenever you want. The interaction flows naturally, matching the rhythm of real-life banter without awkward lag.

Visually, the avatar is far from a frozen picture. As Muse speaks and listens, it displays subtle facial expressions and gestures that mirror the emotional tone of the discussion. You can also customize the setup to your preference: stick to voice only, pin the avatar to the side while you work, or go full screen for a face-to-face chat.

Meta also packed in hundreds of voice personalities. Whether you want the brisk pace of a New Yorker or the dramatic flair of a pirate, you can swap vocal styles on the fly.

Brainstorming without prompt engineering

Meta pitches the tool as a way to work through half-baked thoughts out loud.

Instead of formulating a rigid prompt for a vacation itinerary, you can just mention a random craving or a city you have always wanted to see while cooking dinner or taking a walk. Muse stitches the plan together as you talk.

It is equally handy for role-playing. Users can rehearse job interviews, practice conference presentations, or polish pitches with an interactive partner that answers back in real time.

The feature is currently rolling out to Muse users in the US. Independent tracker TestingCatalog also noted that Muse has begun launching in Mexico. Meta confirmed that hands-free voice conversations will expand to AI Glasses and the Muse Charm wearable in the near future.

Why it matters

We think this is a clever pivot away from prompt engineering toward natural human instinct.

For everyday users and creators, the real bottleneck with generative AI has never been intelligence; it has been the friction of typing structured text. Giving the AI an immediate voice and the ability to handle spoken interruptions makes it far more practical for brainstorming on the move. If you have access, it is easily worth testing for interview prep alone.

For the broader industry, Meta is drawing a clear line against text-heavy chatbots. While OpenAI focuses heavily on audio realism in its Advanced Voice mode, Meta is leveraging its metaverse avatar tech to give the voice a visual face and physical presence.

More importantly, this is a dry run for hardware. Meta explicitly noted that voice is coming to AI Glasses and Muse Charm. Wearable devices do not have keyboards or traditional screens, making fast, interruptible voice conversations the primary operating system. Meta is stress-testing that interface on phones and laptops today so its glasses feel intuitive tomorrow.

The main risk is the uncanny valley. If expressive gestures fall out of sync with complex reasoning, users may quickly revert to voice-only mode. But as a glimpse into how we will interact with ambient AI assistants, Meta just took a remarkably solid step.