Overcomplete ICA algorithm of speech signal extraction in underdetermined mixtures

摘要

An overcomplete ICA algorithm was presented based on the geometric algorithm and shortest-path algorithm for underdetermined blind source separation, i.e. observed signal numbers are less than sources numbers. The algorithm is used to extract speech signal. In speech signals processing, speech signals are collected by microphones, and then use algorithms to extract and separate, when the number of speakers more than the number of microphones. Experimental results indicate that the proposed method has good effect.