Computer Science Program

Kunio OISHI

Name

Kunio OISHI

Department/Research Field

Signal Processing


Research content

The purpose of blind source separation (BSS) is to maximize output signal-to-interference ratio (SIR) for recovering original speech source signals from observed signals without knowing a convolutive mixing process in a noisy and reverberant environment. The approaches for BSS are applicable to direction-of-arrival (DOA) estimation, blind channel identification, hearing aids equipment, and speech enhancement.

Recently, completely new loudspeaker constructions driven by piezoelectric ultrasonic motors or actuators have been developed. The use of an electromechanical transducer with large mechanical impedance for a driver is useful for driving a sound-radiating diaphragm by a strong, as well as highly controlled force. A direct-radiator loudspeaker with this driver works without any significant resonance. It is suitable for very low-frequency signal radiation. Active noise control can be used for the suppression of noise and vibration. It is shown that the filtered-X least mean square (LMS) adaptive filter with the loudspeaker systems driven by piezoelectric ultrasonic actuators is effective to reduce low-frequency refrigerator noise that affects sleep disorder.


Research theme

1. Blind source separation (BSS)
2. Loudspeaker systems with piezoelectric ultrasonic actuators

Research keywords
Convolutive linear mixture , piezoelectric ultrasonic actuator , loudspeaker
Lab link