Pemanfaatan Jaringan Syaraf Tiruan untuk Memprediksi Frekuensi Bunyi pada Bilah Gamelan Saron

Abstract

Gamelan saron is one of the traditional musical instruments from Central Java that is played by beating using panakol saron. Traditionally, Javanese gamelan is tuned based on a subjective sense of harmony, making it necessary to adopt a more objective method for determining its sound frequency. This study aims to identify and predict the fundamental frequency of saron bars using an Artificial Neural Network (ANN) with the backpropagation algorithm. Frequency data were obtained using Audacity and processed in Matlab with a 50-10-1 network architecture and three types of training functions (TrainGD, TrainGDA, and TrainGDM). The best result was achieved using TrainGDA, with a testing MSE of 0.000893 and an accuracy range of 100%–102%. These findings demonstrate that ANN is effective in recognizing saron frequency patterns and supports gamelan preservation through digitalization.