A Stacked Technique for Gender Recognition Through Voice

Detecting the gender of a person (male or female) through their voice seems to be a very trivial task for humans. Our minds are trained over the course of time to detect the differences in voices of males and females. Our ears work as the front end, receiving the audio signals which our brain processes and makes the decision. But it is a challenging problem for computers. Gender classification has applications like, it is able to improve the intelligence of a surveillance system, analyze the customer's demands for store management, and allow the robots to perceive gender etc. This paper proposes a stacked machine learning algorithm to determine gender using the acoustic parameters of voice sample and compares its performance with existing classifiers as CART, Random forest and neural network.