Sounds Prediction Instruments Based Using K-Means and Bat Algorithm

This study focuses on predicting sound categories in music datasets, employing classification to match predicted sounds with correct instrument categories. Sound classification in music datasets is challenging due to factors such as sampling and digital representation. Discretization at a sufficiently high rate helps preserve the original signal than ensures that classify all relevant information to appropriate group. This study proposes a novel approach that integrates K-Means clustering and the Bat Algorithm for efficient data discretization, aiming to improve sound category prediction accuracy. Unlike existing methods, our approach optimizes the data split points, leading to more informative and well-separated groups for classification. We tested our method with two classifiers, Naive Bayes and K-Nearest Neighbors, and observed significant improvements in classification performance compared to using raw data or other discretization methods. This approach offers a promising way to enhance the accuracy of sound prediction in music applications.