Penerapan Metode Mel Frequency Cepstral Coefficients untuk Ekstraksi Fitur pada Deteksi Hukum Pembacaan Al Quran

Abstract

The application to learn the laws of Quran recitation is one solution for Muslims who intend to learn and improve their Quran recitation, but are hindered by limited time and availability of recitation teachers. One of the services that can be offered is a feature to detect whether the reading of the Quran from the user is in accordance with applicable reading laws. This research was carried out to create a model of extraction of ghunnah and ikhfa reading features using the Mel Frequency Cepstral Coefficients (MFCC) method and was carried out through several stages, namely the pre-emphasis, frame blocking, windowing, and the Fast Fourier Transform (FFT) stages. The data used in this study  is the voice of someone recite Q.S. Al-Kahf. The recorded voice data is classified using a neural network perceptron by dividing the data into classes.  The research was conducted using Matlab and testing was carried out based on the calculation of the MAE (Mean Average Error) value. Based on the test results, a MAE value of 0,0062  was obtained.