Christopher W. Turner
Dept. of Commun. Sci. and Disord., Syracuse Univ., 805 S. Crouse Ave., Syracuse, NY 13244
While it has been long known that vowels are, in general, more intense than consonants in naturally spoken speech, this statement is based upon the overall sound levels of the phonemes. The purpose of the present study was to determine the sound levels of consonants and vowels within individual frequency bands, with potential application to the provision of consonant audibility to hearing-impaired patients via amplification. The NST speech materials were digitized and subjected to 1/3 octave-band analysis in contiguous 23-ms samples. The samples were then classified as vowels or consonants, based upon waveform and spectral analysis. A large majority of the samples were vowels, reflecting their greater duration. The long-term consonant spectrum was flat, in contrast to the more familiar low-pass shape of the traditional speech spectrum. In the higher-frequency bands, consonant levels were in many cases more intense than the vowels. [Work supported by NIDCD Grant No. 00377.]