Performance improvement in automatic gender identification using hierarchical clustering

In this paper a hierarchical structure is proposed for automatic gender identification (AGI). In this structure two clustering techniques are used. The first technique is divisive clustering for dividing speakers from each gender to some classes of speakers. The second clustering technique is agglomerative clustering for creating a hierarchical structure. Feature reduction is done by SOAP feature selection to reduce clustering and classification run time. Another idea used in this paper is to choose and use the best classifier in each level of the proposed hierarchical gender classifier. Three classifiers including GMM, SVM and MLP are used. 96.19% gender classification accuracy was obtained on OGI Multilanguage Telephony Corpus which is 3.5%, 2.5%, and 1% better than the performance obtained by traditional classifiers using GMM, SVM and MLP respectively.