A study of REL as an objective evaluation of hypernasality

Objectives: To find an objective index to describe the severity of hypernasality in cleft palate patients through speech spectroanalysis. Methods: 60 audio records of cleft palate patients who went to our speech pathology clinic were analyzed by spectroanalysis. All patients were asked to pronounce the vowel /i/. In the LPC spectrum, the ratio (REL) of the average loudness in the low frequency area (the nasal formant area) to the peak of loudness in the first formant was calculated of every record. One experienced speech therapist and two non-professionals made perceptual judgments of all records, and grades from 0-3 were given according to the severity. The speech pathologist's grades and the average of the two non-professionals grades were statistically analyzed by the Spearman correlation analysis respectively to test if they have significant correlation with the REL data. Results: REL has significant correlation with professional judgments (P<0.05) and non-professional judgments (P<0.05) respectively. Conclusions: REL can be used as an objective index to judge the severity of hypernasality.