A moving sound source tracking based on two successive algorithms

Abstract

In this paper, we proposed a novel method for single moving sound source tracking based on two successive algorithms. Although the MUSIC(MUltiple SIgnal Classification) method is well-known as one of the best high resolution methods for the unmoving source, two serious difficulties occur regarding as computational costs to solve the problem when the MUSIC method is applied in such a situation. The first occurs at the signal subspace tracking, which requires the eigen-decomposition and thus needs the enormous computations. The second occurs at a peak search of the MUSIC spectrum function, which involve many complex multiplying operations. In the proposed method, the PAST algorithm is applied to track sequentially the signal subspace without the eigen-decomposition, and the IPLS algorithm is applied to search successively the peak of MUSIC spectrum. Several experimental results in actual room environments are shown to present that more accurate results equivalent to the MUSIC method can be obtained by the proposed method less computational costs than the MUSIC method.