Objective: To study the spectrum features of nasalized vowels and set up effective indexes to reflect the features.
Methods: The acoustic features of 100 children's speech, aged 5 to 12 years old, consisted of 40 unrepaired cleft palate children, 28 VPI children and 32 normal children, were studied by the application of digital spectrograph.
Results: No significant changes in the frequency of inherent formants of each vowel, appearance of one relatively fixed extra formant in the low frequency area, and the energy of high frequency area being damped obviously were the main spectrum features of nasalized vowels.
Conclusion: There are three main features on the spectrums of nasalized vowels compared to that of oral vowels. Nasal resonance index (NRI) can effectively distinguish the extra formant in low frequency area of oral vowel /a/ from nasalized vowel /a/. Energy damping index (EDI) can effectively reflect that the energy of high frequency area of each nasalized vowel is damped obviously.