Microsoft's VASA-1 AI: Crafting Hyper-Realistic Avatars, Ethical Use Prioritized
April 20, 2024
Microsoft Research Asia has created 'VASA-1', an AI that can produce realistic avatar videos from a single image and audio clip.
VASA-1 specializes in true-to-life facial animations, including lip-sync, expressions, and head movements, with high-resolution and frame rate capabilities.
The AI allows user customization of avatars and supports animation and lip-syncing in multiple languages.
Despite its versatility, VASA-1 struggles with animating the torso and has issues with hair, clothing, and texture fidelity.
While the technology has applications in education, accessibility, and therapy, Microsoft has withheld its release due to ethical concerns and the risk of creating deepfakes.
Legal measures like the DEFIANCE Act are being considered to combat the unauthorized use of deepfake technology.
Summary based on 16 sources
Get a daily email with more Tech stories
Sources

heise online • Apr 19, 2024
VASA: Microsoft AI generates "talking face" from photo and voice recording
PetaPixel • Apr 19, 2024
Microsoft's VASA-1 AI Can Make Any Person's Image Move and Speak
ReadWrite • Apr 19, 2024
Microsoft's new VASA-1 AI model can turn photos into 'talking faces'
The Register • Apr 19, 2024
Microsoft teases deepfake AI that's too good to release