Akademik Veri Yönetim Sistemi
ENGINEERING SCIENCE AND TECHNOLOGY-AN INTERNATIONAL JOURNAL-JESTECH, cilt.36, ss.1-11, 2022 (SCI-Expanded)
Automatic voice pathology detection systems can effectively help clinicians by enabling objective assessment and diagnosis in early stage of voice pathologies. This paper suggests a novel multi-modal architecture utilizing speech and electroglottography (EGG) signals and investigates their effectiveness in automatic detection of voice pathology. The proposed multi-modal framework combines two parallel Convolutional Neural Networks (CNNs), one for voice signals and the other for EGG signals, to obtain deep features. Classical handcrafted features are also obtained in the same manner. These features are then concatenated to obtain a more prominent feature set. In addition, a feature selection method is applied to remove redundant features. Finally, a SVM classifier is utilized to detect the voice pathology. In order to measure the performance of the proposed pathology detection system, various experiments are conducted on Saarbruecken Voice Database (SVD) without excluding any available pathology or sample. The experimental results show that the proposed voice pathology detection method achieves accuracy up to 90.10% using all speech and EGG samples. Also, sensitivity, specificity and F1-score results of 92.9%, 84.6% and 92.57% are obtained, respectively. The proposed method provides better performance than those given in the literature using all SVD samples through cross-validation testing. Hence, it is promising for automatic detection applications of voice pathology.