Laser to help headsets with speech recognition

An Israeli company is developing a small laser that can initially be used for headsets and helmets to assist with speech recognition. A sensor with the laser measures tiny vibrations of the skin, which can be used to reduce errors in speech recognition.

Thanks to the laser, the VocalZoom sensor can detect vibrations in the skin, from the eyes to the throat and neck, and works up to a meter away. Those vibrations are converted into an audio signal that is not disturbed by ambient noise. The company’s CEO and founder, Tal Bakish, tells Technology Review that the sensor can assist traditional voice recognition technology with microphones to reduce the margin of error by 60 to 80 percent.

In August, a first prototype headset from the Chinese speech recognition company iFlytek with the technology should appear. Bakish expects the technology to initially end up in motorcycle helmets and headsets of company employees, but he also says he is working with car manufacturers for voice control in vehicles. An obstacle here is that motorists move their heads more and there are fewer possibilities to place the laser sensor close to the face.