Harmonic sound stream segregation using localization and its application to speech stream segregation

Sound stream segregation is essential to understand auditory events in the real world. In this paper, we present a new method of segregating a series of harmonic sounds. The harmonic structure and sound source direction are used as clues for segregation. The direction information of the sources is used to extract fundamental frequencies of individual harmonic sounds, and harmonic sounds are segregated according to the extracted fundamental frequencies. Sequential grouping of harmonic sounds is achieved by using both sound source directions and fundamental frequencies. An application of the harmonic stream segregation to speech stream segregation is presented. It provides effective speech stream segregation using binaural microphones. Experimental results show that the method reduces the spectrum distortions and the fundamental frequency errors compared to an existing monaural system, and that it can segregate three simultaneous harmonic streams with only two microphones.