Speech Enhancement Algorithm Based on Independent Component Analysis

Abstract

Speech enhancement algorithm based on independent component analysis is proposed. The standard independent component analysis algorithm require that the number of sensors is more than or equal to that of sources, so it is impossible to apply independent component analysis to single channel speech enhancement directly. In this paper, a single channel speech enhancement algorithm is proposed by constructing a noise signal to as another observation signal for single channel speech noise reduction based on independent component analysis, thereby noise and original speech can be separated through independent component analysis. Simulation result shows that much better denoising effect and signal-noise ratio can be obtained by using this algorithm.