Statistical Model Based Speech Enhancement Systems

This paper deals with the problem of enhancing speech signals which have been degraded by statistically independent quasi-stationary noise. We consider the estimation of the clean speech waveform, and of the parameters of autoregressive (AR) models for the clean speech, given the noisy speech. The two problems are demonstrated to be closely related in the sense that a good solution to one of them can be used for achieving a satisfactory solution for the other. The difficulties in solving these estimation problems are mainly due to the lack of explicit knowledge of the statistics of the clean speech signal and of the noise process.