Computers Can Now Read Your Lips (And Put Words In Your Mouth)
Experiments at Disney’s Research Hub have yielded a program that is able to break down the visual elements of speech (visemes) and use them to generate hundreds of possible speech sequences (phonemes) that perfectly synchronize with the video. This work explores the natural ambiguity in visual speech and offers insight for automatic speech recognition and the importance of language modeling.

April 20, 2015 at 01:57PM
via Digg http://ift.tt/1Jnltrq