A priori SNR estimation based on momentum term

In view of the parameters estimation problem of a priori SNR in speech enhancement, a new method in this paper is proposed by adding momentum term. This algorithm maintains the advantages of the direct decision algorithm in effectiveness and briefness, also improves the tracking speed of the instantaneous SNR, therefore it can further suppress the generation of musical noise. Simulation results under several noises settings show that the proposed a priori SNR estimation method based on momentum term has better performance than the direct decision method.