Music Genre Classification and Isolation

Music classification is the classification of music based on various factors like genre, rhythmic structure, harmonic content of the music, duration or instruments played. Classification plays an important role in recommendation systems, organizing audio libraries and discovering trends and preferences. Machine Learning techniques have proved useful in music analysis and to classify music clips into different genres. In this paper, various machine learning algorithms like NN, RNN, CNN, SVC and RFC are implemented and the performance of theses algorithms are measured in terms of accuracy score and listening tests. All the models were trained on the GTZAN dataset. From analyzing the scores, the Random Forest Classifier seems to provide the best performance among the other models.Music source separation is breaking a song into its constituent audio sources, such as the vocals, bass, and drums. It is easy to achieve if it is the original multitrack studio recordings. It is difficult to separate the sources if it’s mixed into one single audio file. Convolutional Neural Network (CNN)-based source separation is tested on the MUSDB dataset in this project. The CNN model creates a binary mask for each source and it is used to create the target-extracted audio file. The performance is measured using accuracy scores and BSS-eval metrics along with listening tests. Bass masking has been found to be the best with this CNN-based masking model and the generated masks for vocals, drums and accompaniment do work but they don’t completely extract the target components.