Normal and Organic Pathology Classification of Female Voices Using SVM Classiifier

In this paper, we propose to achieve the classification of normal and pathologic female voices and essentially the classification between organic female voice pathologies: it’s about edema and nodule pathologies. Besides, we propose to study the effect of the fundamental frequency and the open quotient parameters composing the feature vector on the performance rates in addition to the MFCC and their variations and the energy. In this study, we adopt a two-class SVM classifier and we use the MEEI database. The results show that the open quotient permits to discriminate between the normal and the pathologic voices. However, the fundamental frequency combined with the MFCC and their variations and the energy ensure to distinguish between the diseases.