Speech enhancement based on the EM algorithm

Abstract

In this paper, we proposed the new data enhancement method based on the EM algorithm. Proposed framework uses a probabilistic distribution of describing the noisy data in terms of clean and noise data. It employs a variational approach to find an approximation to the joint posterior distribution which can be used for estimating the clean data. It also allows us to estimate the parameters of the clean and noise data based on the EM algorithm. Through the simulation for speech data and sine wave, we have demonstrated that the suggested method improves the SNR in the noisy environment. The suggested method can be easily applied to various areas of speech enhancement and signal detection.