Artikel
Comparison of automated and perceptual categorization of normal and pathological voices
Suche in Medline nach
Autoren
Veröffentlicht: | 19. April 2011 |
---|
Gliederung
Text
Introduction: The aim of the present study was to evaluate the accuracy of the elaborated automated voice categorization system when classifying voice signal samples into the healthy and pathological classes and to compare it with the classification accuracy attained by human experts.
Methods: Effectiveness of ten different feature sets in the classification of voice recordings of the sustained phonation of the vowel sound /a/ into the healthy and two pathological voice classes was investigated. A new approach to building a sequential committee of support vector machines (SVM) for the classification was proposed. Mathematically based “genetic search” determined the optimal values of hyper-parameters of the committee and the feature sets providing the best performance. Four experienced clinical voice specialists evaluating the same voice recordings served as experts. The “gold standard” for classification was clinically and histologically proven diagnosis.
Results: In the experimental investigations performed using 444 voice recordings of the sustained phonation of the vowel /a/ coming from 148 subjects, three recordings from each subject, we obtained correct classification rate (CCR) of over 92% when classifying into the healthy-pathological voice classes, and over 90% – when classifying into three classes (healthy voice and two nodular or diffuse lesion voice classes). The CCR obtained from human experts was about 74% and 60%, respectively.
Conclusion: When operating on the same experimental conditions, the automated voice discrimination technique based on sequential committee of SVM was considerably more definite than the human experts.