Microsoft patent reveals real-time AI speech rewriting with lip-synced live video
Significant_Mail5398 · reddit · 2026-11-11
A newly surfaced Microsoft patent describes technology that can alter speech and lip movements in real time during live broadcasts: the system flags specific words and replaces them with generative AI while reshaping mouth movements to match.
Key points
- Intended uses include live translation with lip sync, removing profanity, fixing mispronunciations, stripping filler words, and rewriting statements with web-sourced information.
- Notably, it could generate personalized broadcasts where two viewers of the same live event see entirely different versions of what was said.
- The patent raises serious concerns around censorship, misinformation, and whether video can still serve as evidence of what someone actually said.
More from Models
- Muse Ember sweeps Browser Use benchmark Pareto frontier at 3 cents per task — WenhuChen · 2026-11-11
- Interpretability researcher pleads with Anthropic to stop banning him; screenshot shows claude-opus-5-5 — Sauers_ · 2026-11-11
- Models still can't orchestrate themselves, and that's what makes this researcher bearish — iruletheworldmo · 2026-11-11
- OpenCode engineer: Carbon built a Rust UI library overnight and rewrote their desktop app — m4rkmc · 2026-11-11
- Gemini 4 Carbon lands on Databricks day 0, scoring 66.7% on OfficeQA Pro v2 — matei_zaharia · 2026-11-11
- MiniMax H3 open ecosystem mapped: 49 family groups, 63 descendant nodes since August — cocktailpeanut · 2026-11-11