Overview
- Meta rolled out Muse Voice and expressive animated avatars in the United States on Thursday, letting users start real‑time voice or video chats with Muse on mobile, web and Mac by tapping a phone icon.
- The new Muse Realtime Voice and Realtime Avatar technology streams the same speech tokens to generate synchronized audio, lip movement and gestures, supports hundreds of customizable voices, and allows users to interrupt Muse mid‑response.
- Meta also began deploying specialized models trained from its frontier Muse Wonder model, bringing Active Memory for longer contextual recall and Active Focus to recover from task failures to Muse and Meta AI starting Friday.
- Journalists and commentators have raised privacy and transparency questions about how voice data will be stored, whether conversations will train models, and when human contractors might be involved in completing user tasks.
- Meta is staging the rollout US‑first while planning broader device and market expansions, and it aims to bind Muse into workflows with new Spaces for project‑style planners that sync with third‑party apps.