Microsoft Hints At Its Realistic AI-avatar Model 'VASA-1'

Tech giant Microsoft has introduced a new AI model called VASA-1 to create deeply realistic human avatars, according to a media report on Saturday.

The development was indicated by Microsoft’s researchers who claimed their AI model can create an animated video of a talking person with synchronised lip movements, an advanced avatar that can be generated using a mere single image and a speech audio clip. However, the public offering of the model is yet to be disclosed.

VASA can work with artistic photos, songs and non-English speech, researchers added. 

Meanwhile, Microsoft is concerned about the model’s misuse in facilitating deep fake content, as it aims to provide AI responsibly for human well-being.

The company highlighted that there are no plans to release the model’s demo, API, product, implementation details, or any related offerings until it develops confidence about its responsible working under compliance. 

Microsoft’s researchers claimed that the technology can capture a wide spectrum of facial nuances and natural head motions. It is anticipated that the model will be used for providing virtual teachers to students or therapeutic support to people in need.