Hippopotamus Optimization Algorithm with Long Short-Term Memory for Music Genre Classification

Music genre is a traditional method of categorizing music, playing a crucial role in music information retrieval.Music genre classification encompasses various types like Disco, Pop, Jazz, and Rock, with some genres sharing similar characteristics, making the model prone to overlap or misclassification.To address this issue, this research introduces the Hippopotamus Optimization Algorithm with Long Short-Term Memory (HOA-LSTM) for genre classification.The HOA enhances convergence and reduces computational time by effectively exploring and exploiting the search space, thereby improving classification accuracy.LSTM, known for capturing temporal dependencies in audio signals, is resilient to vanishing gradient issues, leading to precise classification.Performance metrics such as accuracy, sensitivity, specificity, Positive Predictive Value (PPV), F-measure, error rate, and computation time are used to evaluate HOA-LSTM.The proposed HOA-LSTM achieves an accuracy of 98.47% and an error rate of 1.53% on the GTZAN dataset, outperforming the Bidirectional Long Short-Term Memory-VGG-16 Network (BiLSTM-VGG-16 Net).