Japanese telco NTT DoCoMo's R&D (research and development) centre has taken the first step towards development of a technology that will allow people to talk on the phone without saying a word, the company said yesterday.
New sensor detects speech through facial movements alone
Engineers are developing a sensor, which detects signals coming from the muscle movements in the cheek and jaw made when people are speaking.
Signals from the sensor are interpreted and the sound being made by the speaker can be determined, but because the system measures such impulses, the user needs to just mouth the words and no actual sound has to be made.
"This technology is still at a basic level," said Mariko Wada, a spokeswoman for NTT DoCoMo in Tokyo. During experiments, engineers have been able to get the system to discern vowel sounds with 100 percent accuracy — a world first according to Wada.
The company hopes to complete the development within five years, Wada said.
This work has been part of NTT DoCoMo's R&D investigation into various styles of human interface. The same R&D centre has in the past unveiled a mobile phone design that used vibrations travelling around the body to deliver an audio signal from a transducer on the wrist to the ear.
As one of the possible applications, the sensor can be attached to a mobile handset and, with the help of a voice synthesiser, mobile-phone users can communicate in silence, Wada said.
Eventually, engineers are also hoping to apply the technology to wireless email services so that typing an email message will be faster and easier.