Abstract: During the present period of cultural integration worldwide, there’s a heredity and innovation conundrum for traditional ethnic vocal arts. The current evaluation system is obviously too subjective and lacks a clear road. There is no clear direction for development. This paper uses an improved BP neural network and LSTM (Long-Short-Term-Memory Network) algorithm to construct a whole evaluation model. To accurately understand the effectiveness and innovativeness of traditional ethnic vocal art inheritance, and to transform the qualitative indicator values into feature vector sets through quantization and to input them into the neural network system so as to achieve an accurate judgment of the status quo of its development and a forecast of the future trends. Collecting multiple ethnic group data and constructing a model database based on traditional multiple school vocal databases. This is compared and analyzed with single-layer BP neural network and classic indicators. From the experimental test results, it can be seen that compared with the single architecture, the proposed integrated architecture has the advantages of faster convergence and higher prediction accuracy. And it also greatly reduces the data bias caused by human factors, which provides scientific basis for the spread and development of ethnic vocal cultural. It also expands the application area of technology for the digital protection of intangible cultural heritage.

Keywords: Jun Ma Neural network algorithm, Multiculturalism, Traditional ethnic vocal music, Inheritance and innovation, Evaluation model

Pages: 370-379