Abstract:
Age and gender classification has extended its number of uses, despite of the rise of social platforms and social media. However, this stands in stark contrast to the large performance increases explained previously for the strongly linked task of audio. In this study, we show that deep convolutional neural networks may be used to learn representations and significantly enhance performance on tasks (CNN). Where we get in the inception the training accuracy was 99%, validation accuracy 97%, testing accuracy 98%. A testing dataset with 1676 audio files related to 39 age and gender categories and 40 epochs is combined into a single section titled "age-gender-test."
|