Penerapan Metode Fast Independent Component Analysis (FastICA) dalam Memisahkan Vokal dan Instrumen Seni Geguntangan

Abstract. Application of Fast Independent Component Analysis (Fastica) Method in Separating Vocals and Instruments in Geguntangan. Gamelan Geguntangan is often used in religious ceremonies to accompany ceremonies and entertain the public. Along with its development, the Geguntangan gamelan is also used to accompany the Pesantian. Geguntangan recording plays instruments and vocal sounds, most of which have been mixed. The mixed sounds caused the learning process to be less effective for people who will study Pesantian. The students could not focus because of the distracting sound of the instrument. This study aims to separate the sound of instruments and vocals of Geguntangan using deflationary-based FastICA. The non-linear function used is Logcosh. This study also examines the effect of mixing matrix variables and alpha values on nonlinear functions on SDR, SIR, and SAR values. The results of the paired t-test carried out by these two values did not have a significant effect on SDR, SIR, and SAR. The difference in the average time of the mixing matrix testing process is 0.09 seconds and 0.42 seconds for testing the alpha value. Keywords: Pesantian, Geguntangan, BSS, FastICA, Deflationary Based.