The pipa, being one of the classical instruments in China and having a distinctive ethnicity, has had a long history of usage in the country. In this paper, the main concern is to improve the expressiveness of music performance of the pipa through the development of strategies which will include organic integration of art and techniques in musical performance. With the acknowledgment of the importance of the ability of performing pipa, in this paper, multi-feature fusion has been used to acquire features of the acoustic signal from playing of pipa, and then wavelet neural network is used to develop a model for pipa performance quality assessment. On this basis, the participants in the study were pipa students studying in the university. Then experiments have been done to compare the level of expressiveness in music before and after implementing the above model. It was discovered from the results of the experiment that the WNN model is highly accurate for evaluation of pipa performance quality with the accuracy being more than 98%. From the result, the musical expressiveness improved by 2.07%, with scores of emotional attitude, process methods, and knowledge skills increasing from 0.237, 0.259, and 0.225 to 0.238, 0.266, and 0.233 respectively.
