Abstract

Age and gender classification has become relevant to an increasing amount of applications, particularly since the rise of social platforms and social media. Even Nevertheless, contrast to the large performance improvements recently reported for the closely related task of audio. In this research, we show that performance on these tasks can be significantly improved by learning representations using deep convolutional neural networks (CNN). where we get in the ResNet the training accuracy was 98% ,validation accuracy 95%, testing accuracy 96% .Testing dataset organized into one folder (age-gender-test) and contains 1676 audio files related to 39 of age and gender categories and 40 epoch