Extraordinary patent from Microsoft: Artificial intelligence will change live conversations

Microsoft has applied for a patent for a system that can replace conversations in live broadcasts with artificial intelligence. The technology can also regulate the speaker's lip movements.

Software giant Microsoft has received a patent for a new artificial intelligence technology that can transform what the speaker says in real time during live broadcasts. The application with registration number US 2026/0341309 A1, brought to light by the Patentlyze platform, was registered in April 2025 and became official on October 29, 2026. The developed system can instantly evaluate the video and audio stream and reconstruct the speech content and related images within seconds.

Live Stream Lip Sync and Filtering

According to the technical details in the patent document, the infrastructure works by taking into account criteria such as the audience's age group, native language and geographical location. The artificial intelligence agent, which scans the sound waves in the live broadcast, can censor inappropriate expressions or replace them with alternative words in line with predetermined parameters. In the visual part of the process, productive models come into play; The speaker's mouth and lip movements are perfectly matched to the newly placed phrases. The mechanism, which works with a minimum delay of approximately 1 to 10 seconds, is aimed to be able to work with local processing power on augmented reality glasses, smartphones, tablets and video cameras.

From Live Translation to Security and Manipulation Concerns

The technology in question promises new possibilities, especially in global broadcasting and simultaneous language translation. While it has become possible to reflect the speaker with lip movements as if he were speaking in a foreign language in live conferences; Stutters, slang expressions, stuttering and mispronunciations during the presentation will also be automatically cleared. On the other hand, experts point out that algorithms that can instantly intervene in live images can take the risk of deepfakes to a different dimension. The ability to change conversations during broadcast can directly affect both the security of personal expression and the accuracy of information in digital media.