Logo image
Back
Perception of spectrally-shifted non-native speech
Abstract   Peer reviewed

Perception of spectrally-shifted non-native speech

Michelle R. Kapolowicz, Daniel Guest, Vahid Montazeri, Melissa M. Baese-Berk and Peter F. Assmann
The Journal of the Acoustical Society of America, Vol.144(3 Suppl), pp.1866-1866
09/2018
DOI: 10.1121/1.5068208

View Online

Abstract

Some studies have suggested that perceiving speech from multiple non-native talkers can be more difficult than perceiving speech from a single non-native talker. We found evidence in favor of this idea, previously showing a detrimental effect on sentence intelligibility of reducing the spectral resolution of non-native speech via a 9-channel tone vocoder (a condition where talker-specific spectral cues are limited). This effect was more severe when perceiving non-native speech from multiple talkers compared to a single talker, whereas we observed no differences across single- and multiple- native talker conditions. These results imply that listeners place greater reliance on talker-specific spectral cues to aid with perception of non-native speech as compared to native speech. In the present study, we further examined the effects of talker-specific spectral cues on intelligibility by scaling the spectral envelope and fundamental frequency (F0) of previously recorded sentences in successive increments of 8% and 30%, respectively, using the STRAIGHT vocoder. As predicted, intelligibility of spectrally-shifted sentences spoken from a single, non-native talker was similar to that observed for multiple non-native talkers, and listeners reported hearing spectrally-shifted speech as coming from different talkers. These results confirm the importance of spectral cues for the perception of non-native speech.

Details

Metrics

1 Record Views
Logo image