Analysis of pathological voices by speech processing