Headrest Application of Multi-Channel Feedback Active Noise Control with Virtual Sensing Technique

Conventional ANC system based on the feedback structure can cancel the periodicity or narrow-band noise around the error microphone. Therefore, the quiet zone is not realized around the eardrums neither opening of ear canal, especially the noise includes high frequency components, because the error microphones are located a bit far from the user's ears. This is related to the zone of quiet (ZoQ), whose size is 1/10 of wavelength of the target noise. A solution to this problem is to apply a virtual sensing technique. This technique can achieve the maximum noise reduction at the desired locations. Currently, two classes of algorithms have been developed for virtual sensing technique. The first class does not require offline training to obtain some system models to compensate the relationship between the physical microphone location and the desired location, but requires a microphone array to estimate the sound pressure and gradient of the sound pressure at the desired locations. However, there are some constraints between the microphones and then this class is unsuitable for the headrest ANC system. We therefore focus on the second class, which requires the system models. In this class, physical microphones are placed at the desired locations and the transfer functions between the error microphones and virtual microphones are estimated in advance. Nevertheless, it is impossible to place physical microphones near the eardrums. Hence, we apply the virtual sensing technique to the headrest ANC system using the implanted microphones of a head and torso simulator. We demonstrate the effectiveness of the headrest ANC system with the virtual sensing technique through some experimental results obtained using a digital signal processor (DSP).