Detection of Gender and Age Category from Speech

The main goal of this paper was to find out how the gender and age group acoustical models behave on audio data that is in no way related to the data corpora used to train and evaluate the models. These models could be used for early predator detection in minors’ audio chats. The datasets used for training were mainly composed of English speakers. One part was made up of speakers whose native language is English and the rest use English as a secondary language.