For the purpose of improving the coding efficiency, this paper attempts to combine psychoacoustic model for perceptual evaluation of audio quality in BS.1387 with perceptual audio coder. The principle of this new psychoacoustic model is analyzed in theory, and corresponding improvements are proposed to make it be effectively applied to actual audio coder. Both this new model and MPEG psychoacoustic model 2 are implemented in the latest AVS reference coder of China, and comparison of output masking parameter and subjective hearing test between the two models are conducted. Experimental results show that the proposed psychoacoustic model is feasible.