Meta Adds Real-Time Voice and Expressive Avatars to Its Muse AI App

The features were previewed at Meta's Connect conference in September. Users start a conversation by tapping a phone icon in the app, and can interrupt the assistant mid-reply, Meta said. The company is pitching voice as a way to work through loosely formed ideas, such as planning a trip, and to rehearse spoken tasks like job interviews, presentations and pitches.

The update matters because Muse is the centre of Meta's push into personal AI agents, and voice is the interface Meta needs for the hardware it is building around the assistant. Meta said voice conversations with Muse are "coming soon" to its AI glasses and to Muse Charm, a small pocket device it showed at Connect that is meant for talking to the agent without opening a phone app. Meta has not given a price or release date for the Charm.

The release relies on two pieces of in-house technology, which Meta calls Muse Realtime Voice and Muse Realtime Avatar. As the assistant speaks, the on-screen avatar uses facial expressions and gestures matched to the tone of the conversation and reacts while the user is talking, the company said. Users can choose voice only, an avatar shown alongside the conversation, or a full-screen face-to-face view.

Meta is also offering what it described as hundreds of voices with different personalities, giving the examples of the energy of a New Yorker or the flair of a pirate. Users can change the voice whenever they like.

The launch puts Muse more directly up against the voice modes already offered by OpenAI's ChatGPT and Google's Gemini, both of which let users hold spoken conversations with their assistants. Meta did not publish usage figures for Muse or say whether the voice features will carry a separate charge.

Voice and avatars are available in the US first, Meta said, without giving a timeline for other countries.