Implementation of audio recognition using mel frequency cepstrum coefficient and dynamic time warping in wirama praharsini

Abstract

Sekar Agung or wirama is a Balinese classic work which contains of moral values, and is usually sung during traditional or religious ceremonies. The classic nature of wirama made this art was abandoned by the younger generation who were less interested to learn or preserve it. Related to the problem, this study aimed at conducting sound matching as a medium to learn wirama praharsini based on the rule of guru and laghu. Wirama Praharsini was chosen because of the simple way of singing and pronunciation when it was compared to the other kinds of wirama. This study applied voice recognition method to identify words' pronunciation in wirama praharsini. MFCC (Mel Frequency Cepstrum Coefficient) was used to extract the voice feature from expert and the tester. The extraction results were compared with the DTW (Dynamic Time Warping) method that was used to compare two sound features from MFCC process. The result of this study showed average accuracy value of 88.89% which indicated that the implementation of MFCC and DTW method could be done to recognize the wirama praharsini pronunciation and detect the right way to sing it based on the rules of guru and laghu in wirama.