Exploiting Non-negative Matrix Factorization with Linear Constraints

This paper investigates a non-negative matrix factorization (NMF)-based approach to the semi-supervised single-channel speech enhancement problem where only non-stationary addi-tive noise signals are given. The proposed method relies on si-nusoidal model of speech production which is integrated inside NMF framework using linear constraints on dictionary atoms. This method is further developed to regularize harmonic ampli-tudes. Simple multiplicative algorithms are presented. The ex-perimental evaluation was made on TIMIT corpus mixed with various types of noise. It has been shown that the proposed method outperforms some of the state-of-the-art noise suppres-sion techniques in terms of signal-to-noise ratio. Index Terms: speech enhancement, sinusoidal model, non-negative matrix factorizatio