Identification of Automated Music Genre by Exploring Machine Learning Approaches

Music genres are some groups of terms generated for the sake of categories the music. Music genres have some typical characteristics. The characteristics are associated with harmonic content of the music, rhythmic structure, and instrumentation. Basically, these Genre hierarchies are utilized to arrange the big collection of songs on the internet. At present this is done manually, so automatic classification is necessary in order to aid the human user. Also, automatic genre classification for music imparts an outline for evaluating and developing attributes for the interpretation of musical signals. For classification purpose different genre classes such as Blues, classical, pop, disco, hip-hop, jazz, reggae, country, metal, and rock. In this proposed work four widely known machine learning algorithms are used to train and test the classifier with a well-known dataset for audio data classification to envisage the genre of music that the audio given as input belongs to. Convolution Neural Network(CNN), Feed-Forward Neural Network(FNN), K- Nearest Neighbours(KNN), Support Vector Machine(SVM) classifiers are used out of which Convolution Neural Network achieved highest accuracy of 81%.