Riskyana Dewi Intan Puspitasari, Atik Wintarti, Elly Matul Imah
Gamelan is an Indonesian cultural heritage that must be preserved and maintained. As a traditional musical instrument made by hand, the tuning is done manually by the master or empu because it requires special skills and usually transmitted from generation to generation, especially for Javanese gamelan. In the tuning process, the empu uses his feeling and hearing sensitivity to adjust the desired gamelan tone. The ability of empu for gamelan tuning needs to be preserved, and possible to do it automatically by machines or computer programs. This study will discuss the automatic gamelan signal tuning process and compare various feature extraction for noise-robust signal recognition on Saron of Gamelan Tone Signal. We used feature extraction methods such as Fast Fourier Transformation (FFT), Principal Component Analysis (PCA), and Wavelet Decomposition with machine learning algorithms to obtain better gamelan signal tuning results. As for the result, the FFT feature gets a consistent result of up to 80 power of noise with an accuracy of 96% compared to other features. © 2022 Elsevier B.V.. All rights reserved.
Data Science Department, Faculty of Mathematics and Science, Universitas Negeri Surabaya, Surabaya, 60213, Indonesia
