Identifying traditional music instruments on polyphonic Indonesian folksong using Mel-Frequency Cepstral Coefficients (MFCC)

In this study, we report our research on identifying the music instruments on polyphonic Indonesian traditional folksongs. We use Mel-Frequency Cepstral Coefficients (MFCC) as a feature extracted from the songs. The music instrument identification is done by applying Support Vector Machine (SVM) algorithm as a classification method on various numbers of instruments and duration of the Indonesian folksongs. The experiment results show that music instrument identification performs better using the classification method. The folksongs that are played using 2 instruments are highly recognized than folksongs that are played using 3 and 4 music instruments. The music instrument identification results are also influenced by the music duration. The music instruments are easily identified in songs that have 2 second duration rather than 4-second and 6-second duration.