Klasifikasi Genre Musik Menggunakan Metode Support Vector Machine Dengan Multi-Kernel

Abstract

Music is a universal art that reflects cultural diversity and individual preferences through various genres. This research explores music genre classification using Support Vector Machine (SVM) with multi-kernel methods. The SVM algorithm, known for its effectiveness in handling complex datasets, is employed to classify music genres based on audio features. The research utilizes the GTZAN dataset, comprising 10 music genres, and extracts audio features from WAV files. After normalization and data splitting, SVM models are trained and evaluated. Results indicate a significant accuracy improvement after hyperparameter tuning, with the best models achieving accuracies of 88.92% for the polynomial kernel and 89.32% for the RBF kernel.