Music Genre Classification using Artificial Neural Network and Spectral Feature Analysis

One of the core tasks in music information retrieval is classifying musical genres, enabling efficient organization, recommendation, and retrieval of music. Traditional classification methods rely on handcrafted features such as timbre, rhythm, and spectral characteristics. However, recent advancements in AI Model, particularly neural networks, have significantly improved classification accuracy. Artificial Neural Networks (ANNs) have been widely employed to learn hierarchical representations directly from raw audio waveforms or spectrograms. This study explores the effectiveness of neural networks in music genre classification, highlighting their ability to capture intricate musical patterns. This work's primary objective is to develop an automatic categorization system by analyzing the various behaviours of musical genres according to their spectral representations. After gathering the appropriately categorized music dataset (GTZAN Music Genre), the extracted feature map is supplied to the neural network model for assessment. Training, testing, and validation accuracy are gained. The performance metrics shows the proposed methodology has highest classification in music genre with an accuracy score of 98.9 1%, precision score of 98.88%, recall score of 98.78%, and F1 score of 98.77%. The significance of rhythmic structure in differentiating musical styles was underlined by the study.