Acoustic Cue Integration in Speech Intonation Recognition With Cochlear Implants

Shu Chen Peng, Monita Chatterjee, Nelson Lu

Research output: Contribution to journalArticle

25 Citations (Scopus)

Abstract

The present article reports on the perceptual weighting of prosodic cues in question-statement identification by adult cochlear implant (CI) listeners. Acoustic analyses of normal-hearing (NH) listeners’ production of sentences spoken as questions or statements confirmed that in English the last bisyllabic word in a sentence carries the dominant cues (F0, duration, and intensity patterns) for the contrast. Furthermore, these analyses showed that the F0 contour is the primary cue for the question-statement contrast, with intensity and duration changes conveying important but less reliable information. On the basis of these acoustic findings, the authors examined adult CI listeners’ performance in two question-statement identification tasks. In Task 1, 13 CI listeners’ question-statement identification accuracy was measured using naturally uttered sentences matched for their syntactic structures. In Task 2, the same listeners’ perceptual cue weighting in question-statement identification was assessed using resynthesized single-word stimuli, within which fundamental frequency (F0), intensity, and duration properties were systematically manipulated. Both tasks were also conducted with four NH listeners with full-spectrum and noise-band-vocoded stimuli. Perceptual cue weighting was assessed by comparing the estimated coefficients in logistic models fitted to the data. Of the 13 CI listeners, 7 achieved high performance levels in Task 1. The results of Task 2 indicated that multiple sources of acoustic cues for question-statement identification were utilized to different extents depending on the listening conditions (e.g., full spectrum vs. spectrally degraded) or the listeners’ hearing and amplification status (e.g., CI vs. NH).

Original languageEnglish (US)
Pages (from-to)67-82
Number of pages16
JournalTrends in Amplification
Volume16
Issue number2
DOIs
StatePublished - Jun 2012

Fingerprint

Cochlear Implants
Acoustics
Cues
Hearing
Noise
Recognition (Psychology)
Logistic Models

Keywords

  • cochlear implants
  • cue weighting
  • intonation
  • perception
  • prosody

ASJC Scopus subject areas

  • Speech and Hearing

Cite this

Acoustic Cue Integration in Speech Intonation Recognition With Cochlear Implants. / Peng, Shu Chen; Chatterjee, Monita; Lu, Nelson.

In: Trends in Amplification, Vol. 16, No. 2, 06.2012, p. 67-82.

Research output: Contribution to journalArticle

@article{92ee0efafadb4424b8d09c252adab4dd,
title = "Acoustic Cue Integration in Speech Intonation Recognition With Cochlear Implants",
abstract = "The present article reports on the perceptual weighting of prosodic cues in question-statement identification by adult cochlear implant (CI) listeners. Acoustic analyses of normal-hearing (NH) listeners’ production of sentences spoken as questions or statements confirmed that in English the last bisyllabic word in a sentence carries the dominant cues (F0, duration, and intensity patterns) for the contrast. Furthermore, these analyses showed that the F0 contour is the primary cue for the question-statement contrast, with intensity and duration changes conveying important but less reliable information. On the basis of these acoustic findings, the authors examined adult CI listeners’ performance in two question-statement identification tasks. In Task 1, 13 CI listeners’ question-statement identification accuracy was measured using naturally uttered sentences matched for their syntactic structures. In Task 2, the same listeners’ perceptual cue weighting in question-statement identification was assessed using resynthesized single-word stimuli, within which fundamental frequency (F0), intensity, and duration properties were systematically manipulated. Both tasks were also conducted with four NH listeners with full-spectrum and noise-band-vocoded stimuli. Perceptual cue weighting was assessed by comparing the estimated coefficients in logistic models fitted to the data. Of the 13 CI listeners, 7 achieved high performance levels in Task 1. The results of Task 2 indicated that multiple sources of acoustic cues for question-statement identification were utilized to different extents depending on the listening conditions (e.g., full spectrum vs. spectrally degraded) or the listeners’ hearing and amplification status (e.g., CI vs. NH).",
keywords = "cochlear implants, cue weighting, intonation, perception, prosody",
author = "Peng, {Shu Chen} and Monita Chatterjee and Nelson Lu",
year = "2012",
month = "6",
doi = "10.1177/1084713812451159",
language = "English (US)",
volume = "16",
pages = "67--82",
journal = "Trends in hearing",
issn = "1084-7138",
publisher = "SAGE Publications Inc.",
number = "2",

}

TY - JOUR

T1 - Acoustic Cue Integration in Speech Intonation Recognition With Cochlear Implants

AU - Peng, Shu Chen

AU - Chatterjee, Monita

AU - Lu, Nelson

PY - 2012/6

Y1 - 2012/6

N2 - The present article reports on the perceptual weighting of prosodic cues in question-statement identification by adult cochlear implant (CI) listeners. Acoustic analyses of normal-hearing (NH) listeners’ production of sentences spoken as questions or statements confirmed that in English the last bisyllabic word in a sentence carries the dominant cues (F0, duration, and intensity patterns) for the contrast. Furthermore, these analyses showed that the F0 contour is the primary cue for the question-statement contrast, with intensity and duration changes conveying important but less reliable information. On the basis of these acoustic findings, the authors examined adult CI listeners’ performance in two question-statement identification tasks. In Task 1, 13 CI listeners’ question-statement identification accuracy was measured using naturally uttered sentences matched for their syntactic structures. In Task 2, the same listeners’ perceptual cue weighting in question-statement identification was assessed using resynthesized single-word stimuli, within which fundamental frequency (F0), intensity, and duration properties were systematically manipulated. Both tasks were also conducted with four NH listeners with full-spectrum and noise-band-vocoded stimuli. Perceptual cue weighting was assessed by comparing the estimated coefficients in logistic models fitted to the data. Of the 13 CI listeners, 7 achieved high performance levels in Task 1. The results of Task 2 indicated that multiple sources of acoustic cues for question-statement identification were utilized to different extents depending on the listening conditions (e.g., full spectrum vs. spectrally degraded) or the listeners’ hearing and amplification status (e.g., CI vs. NH).

AB - The present article reports on the perceptual weighting of prosodic cues in question-statement identification by adult cochlear implant (CI) listeners. Acoustic analyses of normal-hearing (NH) listeners’ production of sentences spoken as questions or statements confirmed that in English the last bisyllabic word in a sentence carries the dominant cues (F0, duration, and intensity patterns) for the contrast. Furthermore, these analyses showed that the F0 contour is the primary cue for the question-statement contrast, with intensity and duration changes conveying important but less reliable information. On the basis of these acoustic findings, the authors examined adult CI listeners’ performance in two question-statement identification tasks. In Task 1, 13 CI listeners’ question-statement identification accuracy was measured using naturally uttered sentences matched for their syntactic structures. In Task 2, the same listeners’ perceptual cue weighting in question-statement identification was assessed using resynthesized single-word stimuli, within which fundamental frequency (F0), intensity, and duration properties were systematically manipulated. Both tasks were also conducted with four NH listeners with full-spectrum and noise-band-vocoded stimuli. Perceptual cue weighting was assessed by comparing the estimated coefficients in logistic models fitted to the data. Of the 13 CI listeners, 7 achieved high performance levels in Task 1. The results of Task 2 indicated that multiple sources of acoustic cues for question-statement identification were utilized to different extents depending on the listening conditions (e.g., full spectrum vs. spectrally degraded) or the listeners’ hearing and amplification status (e.g., CI vs. NH).

KW - cochlear implants

KW - cue weighting

KW - intonation

KW - perception

KW - prosody

UR - http://www.scopus.com/inward/record.url?scp=84867351216&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84867351216&partnerID=8YFLogxK

U2 - 10.1177/1084713812451159

DO - 10.1177/1084713812451159

M3 - Article

C2 - 22790392

AN - SCOPUS:84867351216

VL - 16

SP - 67

EP - 82

JO - Trends in hearing

JF - Trends in hearing

SN - 1084-7138

IS - 2

ER -