Microsoft has filed a patent for an AI-driven system capable of modifying live video content in near real-time. The technology, titled 'Real-time AI-enhanced video or augmented reality live presentation technology,' was submitted in April 2025 and published in October 2026.
The system uses generative AI to monitor audio streams, allowing it to replace, delete, or rewrite spoken content. Simultaneously, it adjusts the speaker's facial expressions and lip movements to match the altered audio, ensuring the final output appears natural to the viewer.
Potential applications include real-time translation with synchronized lip-syncing, the removal of filler words, and the correction of speech errors. The system can also filter out offensive language or dialect-specific terms based on audience preferences, with some configurations using web searches to refine content.
Microsoft suggests the technology could be utilized in lectures, award ceremonies, sports broadcasts, and television programs to reduce manual post-production. To address concerns regarding information integrity, the patent proposes maintaining original audio and video as a 'ground truth' record.
The company also mentions using digital watermarking or similar methods to track modifications. As this is currently only a patent application, there is no guarantee that the technology will be commercialized or integrated into future products.
