M. Matsumoto and S. Hashimoto (Japan)
Blind decomposition, Aggregated microphones, Frequency analysis, The ratio of directivity, Partially disjoint
This paper describes a blind decomposition algorithm uti lizing two directional microphones located at the same place, namely aggregated microphones. Recently, some au thors proposed sound separation algorithm utilizing sparse ness of the sounds. In these methods, we can extract each signal using the time-frequency binary masks. However, these method can not separate the sounds if the sounds overlap in frequency domain due to the sparseness assump tion. In this paper, we treat the problem of separating not only the sounds which hardly overlap but also the sounds which overlap in most of frequency domain. It should be noted that we need to know neither the directivities of two microphones nor the overlapping frequency components in advance. In this paper, we describe the necessary condi tions and the algorithm of the proposed method. Some ex perimental results show that the proposed method can sep arate not only the sounds which hardly overlap but also the sounds which overlap in most of frequency domain.
Important Links:
Go Back
