Microsoft Patents Real-Time AI System That Alters Live Video Speech and Lip Sync

Microsoft has filed a patent for an AI system capable of modifying live video in near real time—detecting and replacing spoken words while editing a speaker's face and mouth to match, creating seamless transitions around any altered audio. The system would operate according to set parameters or user profiles, with described use cases ranging from real-time language translation and filler-word removal to eliminating dialect-specific phrases, mispronunciations, and "offensive" or "inappropriate" words sourced from an internet search.

The patent raises immediate questions about authenticity and media trust. Critics note that automatically tailoring a live speaker's words and appearance to a viewer's personal preferences goes far beyond conventional dubbing, effectively enabling personalized reality filters applied to broadcasts in real time. The patent does acknowledge storing original audio and video as a "ground truth" record—an implicit nod to the potential for misuse. Microsoft did not respond to a request for comment.