Portal de Periódicos da CAPES

An Evaluation of Hoarseness Using Fuzzy Logics.

1993; Japan Broncho-esophagological Society; Volume: 44; Issue: 3 Linguagem: Inglês

10.2468/jbes.44.184

1880-6848

Katsumi Makino, Hiroyuki Fukuda, Tsukasa Sakoh, Akihiro Shiotani, MA Ya, Shunichi Sasaki, Ryouhei Sakaguchi, Yoshihisa Kawasaki, Masahiro Kawaida, Toru Tsuzuki,

Speech Recognition and Synthesis

One way to evaluate hoarseness is through auditory impression. This method can not be performed by just anyone, for it takes an experienced examiner to evaluate the voice according to a set of given voice samples. However, an E. N. T. doctor is able to make a diagnosis of pathological voice quite accurately, with 70-80% probability, by hearing alone, whether or not he has special training in discriminating voice samples. Doctors do this subjectively, on the basis of past experience of the auditory impression of various symptoms and patient data such as age and sex taken into consideration. If there were a method in which a patient could evaluate his own voice in the same way as an experienced doctor without assistance, then such a method could be used as a screening test for voice disorders. We have employed fuzzy logic to calculate numerically such an auditory impression, we here report the development of a diagnostic method for such major disorders as cancer of the larynx (leukoplakia), vocal polyp, polypoid vocal fold, vocal nodule, unilateral laryngeal nerve paralysis and sulcus using fuzzy reasoning results obtained from questionnaires on voice. Using this method, we have been able to achieve approximately 60% accuracy of diagnosis.