Contributions of Consonant-Vowel Transitions to Mandarin Tone Identification in Simulated Electric-Acoustic Hearing

Fei Chen


For hearing-impaired listeners fitted with cochlear implants (CIs), they rely on electric (E) stimulation with primarily slow-varying temporal information but limited spectral information for their speech perception. Many recent studies showed that for those implanted listeners with residual low-frequency hearing, the combined electric-acoustic (E+A) stimulation could significantly improve their speech perception in adverse listening conditions. The present work assessed the contributions of consonant-vowel transitions to Mandarin tone identification via a vocoder based simulation of E+A stimulation. Isolated Mandarin words were processed to preserve full consonants and vowel onsets across consonant-vowel transitions, and replace the rest with noise. The two types of vocoded stimuli, simulating E and E+A stimulations, were presented to normal-hearing Mandarin-speaking listeners to identify lexical tones. Results consistently showed the advantage of E+A stimulation over E-only stimulation when full consonants and the same amount of vowel onset segments were preserved for lexical tone identification. In addition, compared with E stimulation with full vowel segments, the combined-stimulation advantage was observed even when only a small portion of vowel onset segments were presented. Results in this work suggested that in E+A stimulation, segmental contributions were able to provide tone identification benefit relative to E stimulation with the entire Mandarin words.


 DOI: 10.21437/Interspeech.2019-1124

Cite as: Chen, F. (2019) Contributions of Consonant-Vowel Transitions to Mandarin Tone Identification in Simulated Electric-Acoustic Hearing. Proc. Interspeech 2019, 3138-3142, DOI: 10.21437/Interspeech.2019-1124.


@inproceedings{Chen2019,
  author={Fei Chen},
  title={{Contributions of Consonant-Vowel Transitions to Mandarin Tone Identification in Simulated Electric-Acoustic Hearing}},
  year=2019,
  booktitle={Proc. Interspeech 2019},
  pages={3138--3142},
  doi={10.21437/Interspeech.2019-1124},
  url={http://dx.doi.org/10.21437/Interspeech.2019-1124}
}