According to a paper published in Nature Communications, a new speech recognition system developed by
Cochlear implantation is required for patients withbilateral hearing loss or severe deafness. A hearing aid only amplifies the sound. In contrast, a cochlear implant uses a special set of programs to transform and purify sounds, making them more natural.
The authors of the new development wanted to help patients withhearing impaired who cannot always communicate using sign language. However, the device is also useful for ordinary people who are in extreme conditions when radio communications are interfered with by ambient noise.
The device contains strain sensors thatrecord the movements of the skin - compression and expansion when a person pronounces words. Then a deep learning algorithm is connected, which converts facial movements into words.
“A strain sensor attached to the facestretches and contracts when a person speaks. During the process, the electrical properties of the strain gauges change,” Taemin Kim from Yonsei University’s School of Electrical and Electronics Engineering told Euronews.
At the moment the system can recognize the setout of 100 words with almost 88 percent accuracy. As the authors of the development note, the operation of the sensors will not be interfered with by sweat and sebum. The device itself is much smaller than its designed analogues.
Read more:
Named the consequences of a gas leak from the "Nord Stream"
A massive impact from a space object triggered the Earth's magnetic field
Named the main danger of snoring: scientists told how it affects the brain