The Voice Classification Based on Gender Using Backpropagation and K-Means Clustering Algorithm

Abstract

Sound is the identity of all living creature, including Humans. With voice we can do socialize, call people, ask questions, communicate, and even be able to help us to recognize the sex of the person who makes the sound. Nowadays, knowing gender through sound cannot only be done by humans but through a computer. Voice classification using a computer shows increasingly sophisticated technology. Of course this technological advance can also help in terms of security, where the voice can be a key or password in a certain confidentiality. In this study the focus of sound recordings is classified according to the sex of men and women by using the Backpropagation algorithm for training data, then Mel Frequency Cepstral Coefficients (MFCC) will process sound data and get features, and the K-Means Clustering algorithm will classify sound data already processed. The dataset used here is in the form of male and female voice recordings obtained from YouTube videos that have been separated by video sections. There are each 10 male and female voice files for training. As for testing, there are several male and female voice files that are placed in separate folders.