Model That Can Create Realistic Human Videos from Microsoft

Microsoft introduced the VASA-1 model, which can create realistic human videos using a single visual and audio recording. The model is not planned to be widely available for now.

One of the most fascinating features of artificial intelligence technologies is that they can create images and sounds that are indistinguishable from the real thing. Developed by Microsoft researchersVASAThe system named ” is the newest example of this.

The VASA artificial intelligence system can create faces that look like they are actually talking using a single image and audio recording.

VASA-1 can create realistic facial expressions as well as voice

Name of the first model in which the system was used VASA-1. When visual and audio recording is provided to the model, very realistic results emerge. VASA-1; yfacial expressions, fully synchronized lip movements with no delay, and natural head movements can produce.

What the model can do is not limited to matching lips to voice and a few facial expressions. At the same time various emotions, tiny movements on the person’s face that are difficult to notice It can even detect it. This ensures that the results are frighteningly convincing.

Users of VASA-1 will also have control over the videos created. your character gaze direction and distance, and even emotional state. They will be able to change it. One of its most striking features is that it can create results from any type of input. It can create high-resolution video with many different types of data, from artistic photographs to song lyrics to non-English speech.


source site-33