To improve the effect of MMSE-LSA speech enhancement algorithm at low signal-to-noise ratio(SNR), this paper proposes an improved MMSE-LSA algorithm, which uses noncausal estimation for a priori SNR, and introduces a knowledge of speech absence probability so as to modify gain function from MMSE-LSA. Experimental results show that the proposed algorithm is better than traditional MMSE-LSA algorithm in residual noise suppression and SNR improvement.