AI Patent Edits Live Speech and Lip Movement

Microsoft's filing describes near-real-time video editing, but the application does not confirm any product is in development.

Updated on Nov. 2, 2026 in Artificial Intelligence

Bold flat-color editorial illustration of a black camera lens with sinuous sound-wave lines flowing outward against a navy field. — Microsoft has filed a patent application describing an AI system that would edit speech and lip movements in live video, though the company has not confirmed any product. AI Illustration. Upload story photo >

Live Poll

Would you still trust a live video of someone speaking if AI could alter their words?

Microsoft has filed a patent application describing an AI system that modifies speech and facial movements in live video, published as US 2026/0341309 A1. The filing is research-stage intellectual property; it does not confirm Microsoft is developing or launching such a product.

Why it matters

The application sketches a path toward real-time AI editing of how people speak and look on camera, extending generative video techniques into live conversation. It signals where video-call and AR tooling could head, while leaving open whether the design will ever ship.

The filing's only quantified claim is that processing is near-real-time, within a few seconds, with no latency benchmark or comparison to existing tools. No accuracy or performance figures for the speech or video modifications are reported.

The players

Microsoft

Microsoft is a major software and cloud company whose AI research spans generative models, speech, and video technologies.

The details

Cameras and microphones capture live audio and video, and an AI agent analyzes the conversation to flag words to change based on a user's profile, such as removing filler words like 'um' and 'you know,' correcting mispronunciations, or replacing slang. Generative AI, which produces new images from learned patterns, then redraws the speaker's mouth and adjusts surrounding video frames so the altered speech matches the visuals.

Article coverage of the patent application was published on 2026-11-02.

The Tech Race

The patent places Microsoft's research in the field of AI-driven live video editing, where companies are pursuing real-time generative video and speech tools. The filing is an application, not a shipped capability, so its competitive weight depends on whether it moves into product development.

If built, the system would run on smartphones, tablets, cameras, and AR glasses, and could translate a speaker's words into a language not in the viewer's profile. The filing gives no timeline, pricing, or availability, so the practical impact for users remains unannounced.

The takeaway

The patent points to live video conversations being edited in real time, with AI adjusting both words and facial movement. Watch for whether Microsoft announces a product or research demonstration built on this filing, which would be the first verifiable sign the concept is moving from patent to practice.