Modified SAFIA utilizing aggregated microphones

This paper introduces SAFIA: sound source Segregation based on estimating incident Angle of each Frequency component of Input sources Acquired by multiple microphones, which utilizes the level differences of directional microphones located at the same place, namely aggregated microphones. SAFIA can realize sound source separation based on the difference of the power level between two microphones. In the conventional SAFIA, one microphone for target signal (handset) is located near the target signal. The other microphone for noise signal is located far from the target signal. Hence, SAFIA needs the large system size. Our research target is to solve this problem utilizing aggregated microphones. In this paper, we firstly explain the conventional SAFIA system and the problem of SAFIA. We then apply SAFIA to the aggregated microphones and compare the proposed method with the conventional SAFIA. We demonstrate the experimental results of sound separation utilizing the proposed method.