{{Short description|Deep learning model for audio-visual speech recognition}} '''LipNet''' is a deep neural network for audio-visual speech recognition (ASVR). It was created by University of Oxford researchers Yannis Assael, Brendan Shillingford, Shimon Whiteson, and Nando de Freitas.<ref>{{cite arXiv |eprint=1611.01599 |class=cs.LG |first1=Yannis M. |last1=Assael |first2=Brendan |last2=Shillingford |title=LipNet: End-to-End Sentence-level Lipreading |date=2016-12-16 |last3=Whiteson |first3=Shimon |last4=de Freitas |first4=Nando}}</ref> The researchers stated that could match mouth movements to text with 93 percent accuracy,<ref>{{Cite news |date=2016-11-08 |title=AI that lip-reads 'better than humans' |url=https://www.bbc.com/news/technology-37911135 |access-date=2026-05-25 |work=BBC News |language=en-GB}}</ref> though it was criticized for its test using a limited dataset of words and grammar.<ref>{{Cite web|url=https://www.theverge.com/2016/11/7/13551210/ai-deep-learning-lip-reading-accuracy-oxford|title=Can deep learning help solve lip reading?|first=James|last=Vincent|date=November 7, 2016|website=The Verge}}</ref> It was used in Nvidia's autonomous "backseat driver" prototype Co-Pilot.<ref>{{Cite news|url=https://www.theregister.com/2017/01/17/lipreading_ai_nvidia_copilot_self_driving/|title=Revealed: How Nvidia's 'backseat driver' AI learned to read lips|first=Katyanna|last=Quach|website=www.theregister.com}}</ref>
== References == {{reflist}}
Category:Deep learning software applications Category:Artificial neural networks
{{deep-learning-stub}} {{speech-recognition-stub}}