Organic voice pathology classification
Chekili Salma, Asma Belhaj, Bouzid Aicha · 2017
In this paper, we propose to achieve the classification of pathologic voices and essentially the classification between organic pathologies: it's about polyp, edema and nodule pathologies using new features. The principle contribution in this work is to provide new parameter more efficient than the classic MFCC. It's about calculating MFCC not from the speech signal but from the speech multiscale product. In this study, we adopt a three-class SVM classifier and we use the MEEI database. The results show that the classification rates obtained using feature extracted from the multiscale product give better results.