Market analysis reports, financial performance reports from Microsoft

Microsoft Unveils VASA-1 AI Model for Hyper-Realistic Talking Face Videos

Apr 18, 2024, 12:27 AM

Microsoft Research has unveiled a new AI model named VASA-1, which can generate hyper-realistic talking face videos from just a single portrait photo and a speech audio clip. The model produces videos with precise lip-audio synchronization, lifelike facial behavior, and naturalistic head movements. This advancement represents a significant leap in the quality of AI-generated video, delivering photorealistic, emotive talking heads that can lip-sync and exhibit realistic facial expressions in real time.