Pengenalan Chord pada Gitar dengan MFCC Sebagai Metode Ekstraksi Ciri dan Jaringan Saraf Tiruan Sebagai Metode Pengenalan Pola

Abstract

Human hearing system is capable of identifying sounds or audio signals, especially sounds that are familiar in their daily lives. However, recognizing chord sequences played in some kind of music is not an easy task. People need big effort to train their sense of hearing so that they can recognize chords. This condition is also valid for a computer system. Finding the key and labeling the chords automatically from music are of great use for those who want to do harmonic analysis of music. This research is about to recognize chords played and recorded by a guitar instrument. There are 24 chords used in this research. MFCC was used as a feature extraction using 13 and 26 cepstral coefficients. Each chord signal which has been extracted is modeled using artificial neural networks as a method of pattern recognition. This research results in an accuracy level above 90%. From the research that has been done, it can be concluded that modeling using back propagation neural networks on guitar chords recognition has an accuracy of 92%, better than the codebook method performed in the previous research which resulted in an accuracy of 88%. An increasing accuracy level is shown by using artificial neural networks for pattern recognition