This paper considers a speech enhancement based on the psychoacoustically constrained optimization problem. In this paper, four constrained optimization problems are set using the masking threshold of the clean speech, and these problems and their solutions are considered. Experimental results show that the performance of the algorithms considered in this paper.