Pemodelan Support Vector Machine untuk Pengenalan Chord pada Alat Musik Gitar Menggunakan Metode MFCC sebagai Ekstraksi Ciri

Abstract

Various methods of feature extraction and classification methods have been developed by researchers. Application of MFCC feature extraction methods and SVM classification methods will be used in this research to identify the chord on the guitar instrument. The first step is extracting the characteristics of recorded chords. As many as 24 chords were used, with a composition of 75% for training data and 25% for the test data. This research uses SVM with three kernel types: linear, polynomial, and RBF. Each kernel has different parameters. The parameters used are the best parameter resulted from the cross-validation process. The final result showed that the RBF kernel and polynomial has the same highest accuracy of 93.33%.