Meta Launches Real-Time Voice Mode and Avatars for Muse AI

Meta has launched real-time conversational voice capabilities and animated expressive avatars for its consumer artificial intelligence assistant, Muse.

Real-Time Audio Models and Expressive Avatar Animation

First showcased at the company's Connect developer conference in September, the video and voice interaction framework is powered by proprietary Muse Realtime Avatar and Muse Realtime Voice models. The technology enables digital characters to generate synchronized facial expressions and subtle physical gestures that adapt dynamically to the tone and tempo of dialogue. While rival platforms like OpenAI's ChatGPT offer voice chat modes, Meta differentiates its service by pairing audio responses with animated visual avatars, including a default character nicknamed Jolly. Users can personalize the visual identity of their assistant across diverse humanoid, animal, or stylized forms, while selecting varied vocal tones through application settings.

As patrons converse with the assistant, the avatar positioned at the top of the display animates directly in time with spoken words. Listeners can alter the auditory tone of the voice by asking their agent directly or by browsing through available voice options in account settings.

Cross-Device Availability and Hardware Integration

The live conversational features are currently accessible to consumers across the United States on mobile operating systems, web browsers, and macOS desktop environments. Meta stated that voice interactions will expand to its connected smart glasses and a forthcoming Tamagotchi-like wearable hardware device called the Muse Charm before the end of the year. When the dedicated wearable hardware arrives, the digital avatar will render on a compact screen, enabling users to speak with the agent and delegate tasks.

Meta demonstrated the conversational system through a video published on social media platform X by Chief AI Officer Alexandr Wang, who interacted with an avatar styled as a fluffy white dog.

Strategic Context and Infrastructure Investment

To support extensive compute requirements across its consumer AI ecosystem, Meta recently outlined plans for a $25 billion bond offering to fund long-term data center and computing infrastructure.

Background

Amazon blocked Muse from shopping on its site over credential security concerns and lack of agent identification, demanding that agentic applications operate openly and respect service provider participation rules, according to Mashable.