Microsoft Files Patent Application for AI That Could Rewrite Live Speech and Resync a Speaker’s Lips
Reported by Pasquale Pillitteri and Gadget Review2 sources ↓
Microsoft’s published patent application describes AI that edits words in a live audio stream and changes a speaker’s mouth movements to match. It is a proposal, not a product launch; viewer disclosure requirements are unclear.
Key points
- The application was filed on April 29, 2025, and published on October 29, 2026
- Proposed uses include live translation, profanity filtering and filler-word removal
- The filing says original, unedited audio and video would be preserved for review
- Microsoft has not confirmed plans to release the system
Sources
Details
Gadget Review reports that Microsoft Technology Licensing’s application describes AI agents that identify words or phrases for editing in a live stream. A generative AI system would replace or remove the words, smooth the audio transition and alter video frames so the speaker’s mouth movements match the edited audio. The filing calls the processing “near real-time.”
The application lists live translation, profanity filtering, filler-word removal and content customization as possible uses. It also describes changing stutters, mispronunciations and regional slang; in some configurations, replacement wording could come from an internet search. Gadget Review says Patentlyze interprets an example in the filing as a processing window of roughly one to ten seconds, not a measured performance result.
The filing says original audio and video would be retained as a record for review, but reporting does not establish whether viewers would be required to receive visible notice that a stream was altered. Gadget Review reports no product announcement or launch timeline, and says Microsoft has not confirmed plans to ship the proposed system.
Written from coverage as of
Images
- Gadget Review
- Gadget Review
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh