support vector machine; gamelan; music transcription; RELIEF-F; features selection
In this paper we describe an approach of features selection for Javanese gamelan instrument
identification using RELIEF-F algorithm and support vector machines (SVM). The model extracts
the pitch of the signal using Short Time Fourier Transfrom (STFT). Based on the pitch estimation,
the features from gamelan sound were created. Before analyzing in SVM, the priority of the
features were determined using RELIEF-F algorithm. We compared the results obtained on the
feature sets calculated by the RELIEF-F algorithm with the results obtained from another algorithm.
The experiment shows that RELIEF-F algorithm gives a slightly good result than infoGain or