A lightweight sensing method of tooth-touch sound for disabled person using remote controller

Abstract

To support disabled people to use remote controllers, several biological signals are used. The tooth-touch is one of desirable biological signals. This is because it is the simple and natural human behavior. However, a sophisticated signal processing to extract only the tooth-touch sound is needed since the tooth-touch sound is mixed with the voice sound. This paper proposes a lightweight sensing method extracting the tooth-touch without a sophisticated signal processing to eliminate only voice sound from the sound wave in which tooth-touch and voice are mixed. Instead of the audible sound wave, proposal uses a shock wave (i.e., ultrasonic wave) which is generated when the upper tooth and the lower tooth hit each other. Using the shock wave generated to detect the tooth-touch, a trivial high-pass filter can eliminate only voice sound in the lower frequency domain than in the ultrasonic domain including the tooth-touch. Through a preliminary experiment that uses a conventional microphone and well-known digital high-pass filter, we show that the used electret microphone can sense the tooth-touch as ultrasonic wave and the high-pass filter can extract only tooth-touch. Then, we show some robustness of our method by using the sound waves including the voice and tooth-touch. In addition, we design the filtering hardware to implement a small and cheap system-on-chip achieving a real-time operation. Through the implementation of Field Programmable Gate Array, and the simulation, we show that our hardware is small and performs well for a real-time operation.