Microsoft Patents AI That Rewrites Live Speech Per Viewer and Syncs Lip Movements

ZeroStateReflex · x · 2026-11-09

Microsoft has filed a patent for AI-powered live/video post-processing: the system recognizes spoken words in a livestream in real time and uses generative AI to rephrase them based on viewer profiles (age, region, language)—useful for stripping filler words, bleeping profanity, or swapping out specific terms. The patent diagrams show a pipeline that generates replacement scripts per profile, creates audio transitions, and alters the video so the speaker's lip movements match the substituted words, making the changes undetectable to viewers. It's only a patent filing for now and doesn't mean it will ship in a product, but it has already sparked controversy over livestream content being covertly altered.

Original post →

More from Multimodal

Multimodal channel →