Pengenalan Suara Berdasarkan Usia dan Jenis Kelamin Menggunakan Algoritme Support Vector Machine (SVM)

Abstract

Voice recognition using computers is made possible due to the unique characteristics and frequencies that each voice possesses. These vocal characteristics can be grouped according to gender (male or female) or age (child, teenager, or adult). The purpose of this research is to identify individual voices that have been placed within one of the groups based on age and gender. The data used comes from the research of Fransiswa (2010) and consists of six classes: boys (AL), girls (AP), teenage boys (RL), teenage girls (RP), men (TL), and women (TP). The method used to extract vocal characteristics was the Mel Frequency Cepstrum Coefficient (MFCC) and pattern recognition was performed using a Support Vector Machine (SVM) with several variations of the kernel such as Linear, Polynomial, and Radial Basis Function (RBF). The parameters affecting the MFCC process include: The value of the coefficients, overlap, time frame, and the rate of sampling. The average highest accuracy value of 98,24% was obtained using the RBF Kernel