Recursive estimation based on the trended hidden Markov model in speech enhancement

Abstract

In this study, we propose a new speech enhancement based on trended HMM. The trended HMM is a nonstationary state HMM to model the nonstationary speech. The proposed is recursive method based on frame-by-frame using Kalman filter with LP parameters controlled by Markov switching sequence. Experimental results clearly show the superior performance of the proposed method over the standard HMM based method.