TLDR: Microsoft teases the development of a lifelike avatar AI technology called VASA-1 that can create realistic human avatars for conversations, but no release date has been provided due to concerns about deep fake content.
Microsoft researchers have developed an AI model called VASA-1 that can create lifelike human avatars capable of engaging in conversations with nuanced facial expressions. The technology allows for the creation of animated videos using just a single image and a speech audio clip. However, Microsoft has not specified a release date for this technology due to concerns about potential misuse for creating deep fake content.
Key elements of the article:
- Microsoft researchers have developed an AI model, VASA-1, that can create lifelike human avatars for conversations.
- The technology can create animated videos with synchronized lip movements using a single image and a speech audio clip.
- Microsoft has not provided a release date for the technology, citing concerns about misuse for deep fake content.
- The technology, VASA-1, is called “Visual Affective Skills” and can capture a wide spectrum of facial nuances and natural head motions.
- Researchers tout potential benefits of the technology for educational and therapeutic purposes, but emphasize responsible use.
Earlier this year, a consultant impersonated Joe Biden using a deep fake voice-cloning tool, highlighting the potential risks of AI misuse in disinformation campaigns. Microsoft’s cautious approach to releasing VASA-1 reflects a broader trend in the industry to address the ethical challenges of AI development.