Automatic Classification of Dysphonic Voices

GASTÓN SCHLOTTHAUER; MARÍA EUGENIA TORRES; MARÍA CRISTINA JACKSON MENALDI

WSEAS Transactions on Signal Processing

Año: 2006 vol. 2 p. 1260 - 1267

1790-5052

Spasmodic dysphonia (SD) and muscle tension dysphonia (MTD) are two voice disorders which present similar characteristics. Usually they can be differentiated only by experienced voice clinicians. There are many reasons that support the idea that SD is a neurological disease, requiring surgical treatments or, more usually, laryngeal botulinum toxin A injections as a therapeutic option. On the other hand, MTD is a functional disorder correctable with voice therapy. The importance of a correct diagnosis of these two disorders is critical at the treatment selection moment. In the present article, we present the results of an automatic classifier of these voice pathologies that can help the clinicians to confirm their diagnosis. Using only records of the sustained vowel /a/, we extract eight acoustic parameters and apply pattern recognition tools, based on neural networks, in order to classify the voice as normal, SD or MTD. Previous papers on automatic classification between normal and pathological voices do not differentiate between these three groups. In order to evaluate the performance of our classifier in comparison with previous works: a) we separate the voices into normal and pathological (SD and MTD) with the method here proposed and b) we compared with the results obtained using Fisher Linear Discriminant. Our results overcome the best reported separation between normal and pathological voices.