MLPGI: Multilayer Perceptron-Based Gender Identification Over Voice Samples in Supervised Machine Learning

The goal of this proposed work is to design a gender identification system that identifies the gender of the speaker. Gender classification is an emerging area of research for the accomplishment of efficient interaction between human and machine using speech files. Numerous ways have been proposed for the gender classification in the past. Speech recognition serves as a prime approach for the identification of the source. Other means for the gender classification includes gait of person, lips shape, facial recognition, iris code, etc. In this paper, the gender has been classified for machine learning-based systems using speech files. These systems may be deployed for the critical investigations areas like crime scene. There are various challenges in this field of speech recognition like determining the multilingual segments added in the speech stream and the gender of the speaker. To resolve these problems and to identify the gender of the speaker many different algorithms are used like frequency estimation, matrix representation, Gaussian mixture models, pattern matching algorithm, hidden Markov model, vector quantization, decision trees and neural networks. In this work, we used machine learning methods of neural network and decision trees for the classification of gender that are explained further in literature review.