RÉSUMÉ

Acoustical measures of vocal function are important in the assessments of disordered voice, and for monitoring patients' progress over the course of voice therapy. In the last 2 decades, a variety of techniques for automatic pathological voice detection have been proposed, ranging from traditional temporal or spectral approaches to advanced time-frequency techniques. However, comparison of these methods is a difficult task because of the diversity of approaches. In this article, we explain a framework that holds the existing methods. In the light of this framework, the methodologic principles of disordered voice analysis schemes are compared and discussed. In addition, this article presents a comprehensive review to demonstrate the advantages of time-frequency approaches in analyzing and extracting pathological structures from speech signals. This information may have an important role in the development of new approaches to this problem.