<![CDATA[Human voice is a sound produced when speaking, singing, laughing, and crying. In singing, each individual has unique voice characteristics, including variations in style, pitch, and vocal quality. Human voice types are generally divided into soprano, alto, tenor, and bass. However, identifying singers' voices still faces complex challenges. The main challenges include voice variability, overlapping characteristics between singers, and limited datasets. To overcome these, an accurate identification system is needed. This research uses 600 voice samples from 60 singers in the Talitakumi Pasir Panjang Church youth choir, with durations of 2-20 seconds in WAV format. The feature extraction method used is Mel-Frequency Cepstral Coefficients (MFCC), involving signal recording, preprocessing, signal segmentation, and Fourier transformation. For classification, Probabilistic Neural Network (PNN) is chosen due to its ability to generate probability distributions for each class. This research aims to develop a singer voice type classification system using PNN with MFCC feature extraction. It is hoped that this system can assist vocal trainers in classifying singer voice types more accurately, making a significant contribution to the field of voice and music analysis.]]>
Copyrights © 2026
