Based On Deep Learning Music Genre And Chinese Traditional Instrument Recognition Classification Research

In recent years,with the development of the Internet and digital audio technology,Music Information Retrieval(MIR)has become a research hotspot.In this field,the effective classification of music genres is an important research direction.In addition,instrument recognition is also one of the research hotspots.Chinese traditional instrument as an important part of the world musical instrument,also has great research value,but the research about it is less.At present,the recognition and classification system in the field of MIR is mainly to extract the music features manually first,then the classifier is used to train the model,and finally the built model is used to recognize and classify the music.However,bottlenecks are encountered in manually extracting features.It is very difficult to manually extract music features because the music features required by different recognition tasks may be different,and even some of the music features required by the task may not be named.As a new feature extraction technique,deep learning has made great achievements in image processing and Natural Language Understanding.Therefore,this paper researches whether it is possible to use the powerful feature extraction capability of deep learning to find more suitable music features for music genres and Chinese traditional musical instruments recognition and classification.First of all,this paper introduces the commonly used music features and classical classification methods,and outlines the development history,structure and common models of deep learning,and points out the reason why this paper adopts Deep Belief Networks(DBN).Then,this paper researches the music genre recognition and classification algorithm based on DBN,and it has been improved.Firstly,the Mel-Phon Cofficients(MPC)of music are extracted after preprocessing.Then the features are used as the input of DBN,and dropout and momentum are added to optimize the network,and through constant tuning to train the network.Finally,the best model is used to test music genres.The simulation results show that the accuracy of this method is as high as 75.8%,which is much superior to the existing classical algorithms in the recognition and classification of the 10 music genres of GTZAN library.Finally,this paper presents a Chinese traditional musical instrument recognition and classification algorithm based on DBN.The Mel Frequency Cepstral Coefficients(MFCC)of music samples are extracted after preprocessing firstly,and then they are input into the DBN to train the model,finally the trained model is used to detect the type of Chinese traditional musical instrument.The simulation results show that the accuracy of this method is as high ad 99.2%,which is better than that of the classical algorithms in the recognition and classification of the 6 kinds of instruments of the personal established Chinese traditional musical instrument library.The music quality of musical instrument library is better and there are fewer instruments,so the accuracy rate of the musical instrument library is higher than that of the genre library.