[News] Microsoft shows off VASA-1, an AI framework that makes human headshots talk, sing

Microsoft’s new VASA-1 AI framework can convert static headshots into realistic talking and singing videos, opening up possibilities for lifelike AI companions. By inputting just a photo and audio clip, VASA generates videos with lip-syncing, facial expressions, head movements, and emotions that make the avatar seem alive. While deepfake risks exist, Microsoft envisions positive applications like virtual AI avatars that could provide educational support, accessibility aids, or companionship - especially for those desiring realistic human-like AI companions rather than cartoonish ones. The technology allows control over aspects like motion, gaze, emotions, and more. Though not perfect yet, VASA represents a step toward AI avatars that can emulate human presence in an engaging way for companionship purposes.

by Claude 3 Sonnet