Research And Implementation Of Folk Music Classification System Based On Multiple Features

Folk music is a treasure in Chinese traditional culture.It contains cultural characteristics of various places and reflects the core value of Chinese traditional culture.However,since the founding of the country,the development of most of the folk music has stagnated,and some of the songs have even become extinct or endangered.In this kind of dilemma,combining folk music with classification technology,using classification technology to classify a large number of unorganized unlabeled drama files,this will help folk music to better enter the ordinary people to a certain extent Vision.This paper analyzes the characteristics of folk music.Local operas have their own uniqueness in their vocals,rhythms,and accompaniment instruments.These factors are usually very different between different tracks.According to these characteristics,the timbre characteristics are selected: Mel's cepstrum coefficient,and melody characteristics:fundamental frequency,formant,and band energy.These two characteristics are used as the characteristic parameters of folk music.In this paper,two deep learning models,shallow network logistic regression and deep network deep confidence network,are selected respectively.The experimental design of these two models is validated to verify the applicability of the model to folk music classification.The experimental results show that the classification accuracy of the deep confidence network model is higher than that of the logistic regression model,but the number of iterations required for the accuracy to converge is more than that of the logistic regression model.Finally,the paper determines the overall framework of the folk music classification system based on user needs analysis,and designs and implements the functions of each sub-module of the system.