Music Genre Detection Algorithm

This group project involved developing a system to automatically classify music tracks into five genres—classical, hip-hop, jazz, pop, and reggae—using Python. We extracted audio features from music files with the Librosa library, stored the features as JSON, and used pandas for data handling. Genre classification was performed with a logistic regression model from scikit-learn, and performance was evaluated using confusion matrices and classification reports. Key features included MFCCs, tempo, RMS, spectral bandwidth, centroid, contrast, rolloff, and zero-crossing rate. My main contributions were selecting relevant features, visualizing and analyzing the model (including confusion matrix heatmaps and boxplots), and optimizing feature sets to achieve a final model accuracy of 77%.