Automatic Diagnosis of Pathological Voices Using Pattern Recognition

GASTÓN SCHLOTTHAUER; MARÍA EUGENIA TORRES; MARÍA CRISTINA JACKSON MENALDI

Mendoza, Argentina

Jornada; AST 2006, Simposio Argentino de Tecnología, en el marco de la 35 Jornadas Argentinas de Informática e Investigación Operativa (JAIIO'2006),; 2006

Sociedad Argentina de Informática

A well-known problem in the study of voice pathologies is the discrimination between spasmodic dysphonia (SD) and muscle tension dysphonia (MTD). SD is not very common and many physicians are not familiar with it. Therefore, individuals with SD often are not identified for treatment. There is no automatic method for a proper differential diagnosis. As a preliminary approach to the problem, we use standard clinical measurements of speech, based on the registered sustained vowel /a/, and pattern recognition tools in order to achieve an automatic classification of pathological voices. The results overcome the best reported classification between pathological and normal voices, and have a good behaviour for discrimination between  MTD and SD. The final goal of thepresent work is to provide voice pathologists with a new tool to confirm.their diagnosis.