Publication

Abstract : Identifying human personality is one among the emerging research area. This paper addresses personality identification through introvert and extrovert classification using speech samples. A self-recorded datbase is considered. A novel method of feature extraction involves speech feature derived from Auditory Nerve (AN) Modelling of human ear in identifying human personality for the first time. The effect of the features proposed for the task are compared with Interspeech 2010 features. Bayes-Net Classifier is applied for classification. Furthermore, Voice Activity Detection (VAD) and dimensionality reduction using Attribute Selection technique are applied and the system performance is investigated. The best performance is obtained using VAD and Attribute Selection on the chosen combination of speech features with an accuracy of 88.3%.