Laugh-sensing software developed
Laugh-sensing software developed
2008-07-14Newsfeed
Software that can generate facial animation thanks to its ability to recognise "non-linguistic" sounds, such as laughter, could soon be used to animate computer avatars, it has been revealed.
Software that can generate facial animation thanks to its ability to recognise "non-linguistic" sounds, such as laughter, could soon be used to animate computer avatars, it has been revealed.
Researchers at the University of Bath and the University of Cardiff have developed software that can automatically recognise facial expressions linked to non-speech sounds, such as a laugh, cry, yawn or sneeze. It can then translate these into appropriate animation sequences, which have the potential of improving the quality of web-based avatars or computer-animated movies.
By using optical motion capture, researchers Darren Cosker and Cathy Holt measured the facial expressions and associated voices of four volunteers as they laughed, sobbed, sneezed and yawned. They then developed software that could link the key audio features of the participants' voices with the facial motion-capture data and go on to animate a standardised facial model.
"Providing a person laughs with this standard structure, the computer can take their voice and create an animation sequence," Dr Cosker told New Scientist magazine. He added that the technique had some limitations, with the software unable to detect variations such as those between a loud guffaw and a snigger.
"One person's laugh can sound similar to another person's crying," he said. "In terms of classifying actions on the basis of audio alone, we still need to do more work."
Copyright © The Press Association 2008
