Machine lip-reading technology could help lip readers improve their ability to read others’ lips. A University of East Anglia team led by Sarah Hilder contrasted the success rate of a machine lip-reading program with that of human lip readers. Machine readers had a success rate of 80 percent, while humans had a success rate of only 32 percent. Machines also could use an abstract face shape to interpret lips, while humans needed a video of a real person. Researchers gave volunteers with weak lip-reading skills a chance to try a new training program that improved their ability to interpret monosyllabic words in a few hours. The program, which is video-based, enabled users to see moving lips and gestures as opposed to two-dimensional drawings. “With just four hours of training, it helped them improve their lip-reading skills markedly,” Hilder says. “We hope this research will represent a real technological advance for the deaf community.” One possibility would be free online video lessons, says Royal National Institute for Deaf People campaign manager Agnes Hoctor.
For More Information:http://www.cpccci.com

