Music Genre Classification using Artificial Neural Networks

In the realm of music research, the analysis of musical genres holds paramount importance, and the application of artificial neural network models has garnered increasing interest. This comprehensive study delves into the examination of three pivotal models— multilayer-perceptron (MLP), convolutional-neural-network (CNN), and recurrent-neuralnetwork with Long Short-term Memory (RNN-LSTM)—specifically for music genre classification. The motivation behind this exploration lies in the desire to unravel the potential of these neural network models in deciphering the intricate landscape of musical categorization.Employing a substantial audio dataset extracted from diverse musical compositions, our investigation meticulously compares the performance of the aforementioned models. Noteworthy is the finding that while MLPs exhibit utility in specific applications, they grapple with inherent limitations. Contrarily, RNN-LSTM and CNN emerge as frontrunners, surpassing MLPs in terms of accuracy and loss metrics when classifying music genres. Particularly, CNN showcases significant promise in its capacity to enhance precision in the nuanced domain of music analysis.