Performance Analysis of Deep Learning and Machine Learning Methods for Music Genre Classification System

Abstract

Classification plays a crucial role in numerous applications within the music industry, spanning from content management to personalized playlists and music recommendation systems. While previous research has explored various machine learning frameworks for this purpose, such as support vector machines (SVM), a comprehensive comparison analysis of convolutional neural networks (CNN) and k-nearest neighbor (KNN) remains unknown. This study aims to address this gap by analyzing and contrasting the performance of SVM, KNN, and CNN in music genre classification. Each algorithm was carefully trained using different music genres from a protected dataset, employing feature extraction methods to capture the appropriate qualities of audio signals. The models underwent extensive training with a limited number of samples, and their performance was evaluated using industry standards such as accuracy, precision, recall, and F1 scores. Experimental results included SVM, KNN, and CNN for music genre designation. This study contributes significantly to existing literature by providing a comparative analysis of these algorithms. The findings highlight the strengths and limitations of each approach, offering guidance for researchers and practitioners in choosing the most suitable approach for their specific needs. The insights gained from this research have the potential to enhance music genre classification systems, ultimately improving the user experience across various music-related contexts.