Classification of Musical Genres Using Audio Spectrograms

The classification of music into distinct genres is a valuable undertaking in the realm of multimedia study, as it enables artists, albums, and songs to be categorized based on shared musical characteristics. This study aims to contribute to the field of music genre categorization by proposing an approach that can more accurately classify musical genres compared to existing methodologies. Our approach leverages the attributes extracted from the MK2 audio dataset, which is a compilation of musical genre data that has been collected and processed to facilitate the prediction of song genres. The proposed method employs convolutional neural networks (CNN) and utilizes audio spectrograms as input features. Additionally, this study presents a comparative analysis of our proposed approach with previous research papers, providing insights into the effectiveness of our proposed model.