←News detail

cnBetaUpdated Nov 6 04:21

Meta adds real-time voice conversation and dynamic avatars to its Muse AI assistant

English text is being prepared.

AI summary

Meta adds real-time voice conversation and animated avatars to its Muse AI assistant, supporting natural turn-taking and real-time interruption. Users can choose a style from hundreds of voices, or describe speed, tone, and accent in text to let Muse adjust itself. During conversations, the Muse avatar displays corresponding expressions and gestures based on the content, and Meta says the experience is powered by Muse Realtime Voice and Muse Realtime Avatar technology. A Verge reporter found during testing that Muse usually responds quickly, but occasionally pauses when looking up information; the avatar animation performs well but is sometimes slightly eerie, and the voice also cannot fully imitate the Mark Zuckerberg persona the reporter selected.

Why read this

AI score 58

The original text records three capabilities: real-time voice interruption, voice customization, and animated avatars, and also preserves details about response pauses during testing and the somewhat eerie appearance of the avatar.

Topics

Tags

Body

用户可以直接说出还没整理成熟的想法,让Muse边听边协助构思、规划旅行,也可用它模拟面试、演练演讲或准备提案。语音支持自然接话和实时打断,用户还能从数百种声音中挑选风格,或用文字描述想要的语速、语气和口音,让Muse自行调整。

对话过程中,Muse头像会随谈话内容呈现相应表情与手势。Meta表示,这项体验由Muse Realtime Voice和Muse Realtime Avatar技术驱动。The Verge记者试用时发现,Muse通常响应迅速,但在查找资料时偶尔会停顿;头像的动画表现不错,不过有时略显诡异,语音也无法完全模仿记者选择的马克·扎克伯格形象。

Source

cnBeta