“Hey, Siri, how’s the weather today?” Speech recognition technology is becoming more and more widespread. It is a practical technology with wide applications. However, to get the most out of its intended functions, users should stand close to the device and articulate carefully. What if the skin on our body could recognize voices without the use of devices?
Professor Kilwon Cho and Dr. Siyoung Lee from the Department of Chemical Engineering, and Professor Wonkyu Moon and Dr. Junsoo Kim from the Department of Mechanical Engineering at POSTECH have developed a microphone that detects sound by applying polymeric materials to microelectromechanical systems ( MEMS)
The small, thin microphone presents a wider hearing field than human ears, while it can be easily attached to the skin.
Conventional MEMS microphones used in cell phones, Bluetooth devices, and others are made up of thin, small, and sophisticated diaphragm structures. However, because they are made of rigid and brittle silicon, it is difficult to bend the diaphragm or microphone as desired, which interferes with the device’s sound detection capability.
The research team overcame this limitation by creating a MEMS-based microphone structure using polymeric materials that are more flexible than silicon and can be designed into any shape. The size of the device is a quarter of a fingernail and its thickness is only a few hundred micrometers (μm, 1μm = one millionth of a meter). The microphone can be fixed on large areas of the body or even on the finger.
According to the study, the hearing sensitivity of the microphone is higher than that of human ears, while recognizing surrounding sounds and the user’s voice without distortion. Additionally, it can detect both loud sounds over 85 decibels, a range that causes hearing damage, and low-frequency sounds that humans cannot hear.
The quality of voice detection is comparable to that of cell phones or studio microphones. When the acoustic sensor on the skin was connected to a commercial voice assistant program (Google Assistant), the user could search, translate and control devices effortlessly.
The novel acoustic sensor has potential applications in wearable voice recognition devices for the Internet of Things (IoT) and human-machine interfaces. The research team plans to create electronic skin by integrating it with pressure and temperature sensors that can be attached to the skin, flexible screens and more.