Cues for perception of synthetic and natural diphthongs in either noise or reverberation

Abstract

The relationship between relative intensity of transition segments and identification of diphthongs has been investigated. In the first experiment, synthesized stimuli were used. The stimuli differed in the amount of attenuation of the transition segment which ranged from 0 to 15 dB. It was expected that [diphthong in text] responses would be obtained for stimuli with attenuated transitions. The stimuli were tested in quiet, noise, and reverberation with ten normal-hearing and seven hearing-impaired subjects. For the stimulus with the most attenuated transition, the normal-hearing subjects gave no [diphthong in text] responses and the hearing-impaired subjects gave only 20% [diphthong in text] responses in quiet. However, in noise, both groups of subjects gave 70% [diphthong in text] responses and in reverberation, the normal-hearing subjects gave 95% and the hearing-impaired subjects gave 90% [diphthong in text] responses. Generally, less transition attenuation was needed for the hearing-impaired than for the normal-hearing subjects to give [diphthong in text] responses. These findings indicated that identification errors in noise and reverberation for naturally produced diphthongs might be related to the intensity of their transition segments. In the second experiment, naturally produced diphthongs [diphthongs in text] from the Nábĕlek et al. [J. Acoust. Soc. Am. 92, 1228-1246 (1992)] study were spectrally analyzed. There were 30 different tokens for each diphthong. The results of the analyses indicated significant correlations between the number of identification errors for these diphthongs made by either normal-hearing or hearing-impaired subjects and the relative intensities of the F2 transition segment. In both noise and reverberation there were fewer errors for the diphthong tokens characterized by high intensity F2 transitions.