Effect of consonant-vowel intensity ratio on the intelligibility of spectrally degraded speech
Uma Balakrishnan, Richard L. Freyman, Chiang Yuan Chuan, G. Patrick Nerbonne, Kelly J. Shea · The Journal of the Acoustical Society of America · 1991
Normal-hearing subjects' recognition of spectrally degraded speech was evaluated under conditions where the waveform envelope was modified by altering the consonant-vowel intensity (C-V) ratios. Subjects were required to identify 22 consonants presented in the /a-Consonant-a/ format at suprathreshold and near threshold levels of presentation. In both conditions, the stimuli were processed to limit spectral information. In the suprathreshold condition, stimuli were presented at the natural C-V ratio and at six modified C-V ratios. In the near threshold conditions, three C-V ratios were employed—the natural ratio and two altered C-V ratios. The effect of C-V ratio modification on recognition performance varied across consonants. Within groups of consonants, “high” or “low” C-V ratios predisposed listeners to select certain consonants as the response. For example, for high C-V ratios of the stimuli /afa/ and /asa/, listeners selected /asa/ as the response. For low C-V ratios of these consonants, the predominant response was /afa/. Similar C-V ratio-dependent responses were also obtained for the glide and nasal consonant groups. Finally, for some consonants such as the stops, C-V ratio modification did not affect recognition performance. Results from both studies suggest that in the absence of spectral information, listeners depend on the envelope cues available in the speech waveform for speech sound recognition. [Work supported by the Whitaker Foundation.]