Machine learning technique boosts lip-reading accuracy

a technology model for lip-reading developed at the University of East Anglia in the UK has been shown to be able to interpret mouthed words with a greater degree of accuracy than human lip readers, thanks to the application of machine learning tech to classify the visual aspect of sounds. And the kicker is the algorithm doesn’t need to know the context of what you’re discussing to be able to identify the words you’re using.

While the model remains a piece of research at this stage, there are scores of potential applications for technology that could automagically transform visual cues into accurate speech — whether it’s helping people who have audio impairments, or enhancing audio-less security video footage with additional speech data — or even to try to figure out exactly what charged word one footballer spat at another in the heat of a match.

nplus1.ru