Microsoft’s new AI model can turn your photos into talking faces

You’ve heard about images, videos and even GIFs — get ready for an AI-driven solution that will convert images into talking faces. Christened VASA-1, Microsoft has announced an AI model that will be able to generate lifelike expressions only by using a static image and a voice clip.

Microsoft VASA-1: Eerie or groundbreaking?

VASA-1 will essentially spin a static selfie and turn it into a talking clip of you. All you have to do is upload a photo along with a voice note and let the AI model do the talking for you. 

VASA-1 will be able to render a minute-long clip at 512 x 512 pixels resolution, and at up to 40 frames per second, without compromising on the quality of your image. 

ALSO READ: 5 Microsoft Copilot features you should know about

What’s cool is its ability to generate lip-sync perfectly without the final result appearing off. But if that’s not all, it can also take into consideration the eye gaze direction, head distance, emotion offsets, and more. Lastly, the AI model can also take into account non-English speech and create singing videos if required. 

Despite its usefulness, the brand, however, remains cognizant of the possible misuse that goes hand in hand with this. That said, Microsoft doesn’t entirely address how it plans to curb the growing spread of deepfake videos. In a statement, the brand goes on to mention how VASA-1 will primarily be used to improve forgery detection. 

It also recognises the ‘likelihood of its misuse‘ but talks about how the primary aim is only ‘promoting educational fairness to increasing accessibility for those with communication difficulties to providing companionship or therapeutic assistance’. 

While the brand assures that the key usage will centre around research, the thought of it falling into the public domain seems alarming. Alternatively, several speculations also claim that it could also become a part of OpenAI’s Sora, given its sizeable stake there. But, of course, all this is best taken with a pinch of salt. For any other notable announcements, don’t forget to watch this space.