A New Methodical Perspective for Classification and Recognition of Music Genre Using Machine Learning Classifiers

To tackle difficulties such as finding similar songs, identifying cultures that would enjoy certain music, conducting surveys and music therapy, Classification and Recognition of Music Genre is necessary. The expansion of music-related content available online has created new opportunities for the development of music referral systems, which support user communities in searching, finding, sharing, and producing digitally. We develop a new methodical perspective to select the most optimum machine-learning algorithm for predicting song genre using Support Vector Machine (SVM), k-nearest neighbor (k-NN), Probabilistic Neural Network (PNN) and Convolutional Neural Network. We got the results for CNN having system's accuracy level is 98.9% and a precision of 98.7%, f1 score of 97.5%, recall of 98.5%. and the accuracy level of the system is 98.3% from SVM, 98.5% from PNN, 99.9% from KNN.