A Comprehensive Analysis of Music Genre Classification with Audio Spectrograms using Deep Learning Techniques
Mohammad Kazim Abbas, Kunal Gupta, Mohammad Anas Mudassir, Rishabh Jain · 2023
This review paper investigates the application of audio spectrograms for categorization of musical genres. The use of feature extraction technique in convolutional neural networks (CNNs) for music genre analysis is highlighted in the discussion of these techniques. The relationship between audio spectrograms and musical subgenres is also observed in this paper, as well as how well audio segmentation methods perform in terms of increasing classification precision. The findings of recent research employing various approaches like CNN, PRCNN, LSTM, SVM, etc are used to conduct a comparative analysis of their effectiveness, highlighting the strengths and weaknesses of each approach. The use of spectrogram for feature extraction and multi-layer CNN proved to be ideal for future researches.