Effects of real-time cochlear implant simulation on speech production

Abstract

Investigations using normal-hearing subjects listening to simulations of cochlear implant (CI) acoustic processing have provided substantial information about the impact of these distorted listening conditions on the accuracy of auditory perception, but extensions of this method to the domain of speech production have been limited. In the present study, a portable, real-time vocoder was used to simulate conditions of CI auditory feedback during speech production in NH subjects. Acoustic-phonetic characteristics of sibilant fricatives, aspirated stops, and F1/F2 vowel qualities were analyzed for changes as a result of CI simulation of acoustic speech feedback. Significant changes specific to F1 were observed; speakers reduced their phonological vowel height contrast, typically via talker-specific raising of the low vowels [ae] and [A] or lowering of high vowels [i] and [u]. Comparisons to the results of both localized feedback perturbation procedures and investigations of speech production in deaf adults with CIs are discussed.

Supra-Segmental Changes in Speech Production as a Result of Spectral Feedback Degradation: Comparison with Lombard Speech

“…Taken together, the above re-analysis of speech collected in Casserly (2015) shows tentative support for the hypothesized links between CI-simulated feedback effects and Lombard speech in terms of the speaking rate, mean F0, and voice amplitude domains. The majority of similarities disappear after a short perceptual learning period for speakers, however, and the initial reduction in F0 variability is the opposite of what is typically seen in Lombard speech.…”

“…Speech was collected from participants under three conditions: with normal acoustic feedback (Baseline); immediately after introducing a real-time CI-simulator as an acoustic feedback manipulation (Epoch 2); and following 15 minutes of naturalistic experience with CI-simulated acoustics in a one-on-one conversation with a researcher (Epoch 3). The segmental acoustic-phonetics of isolated words produced across those conditions were reported in Casserly (2015), as well as the segmental phonetics from a small group of control talkers' speech.…”

“…All subjects read aloud 114 isolated English words and a set of 12 meaningful, "phoneme-specific" English sentences. Segmental acoustic phonetics were analyzed using stressed vowels in the isolated word data set (described in more detail in Casserly, 2015), but the supra-segmental characteristics of single-word stimuli are limited. We therefore examined rate, F0, and amplitude using the sentences.…”

“…Casserly (2015) tested this prediction, examining domains of segmental acoustic-phonetics that rely on production of precise spectral frequencies: sibilant fricative centroids and vocalic F1/F2 qualities. Neither area showed the predicted compensation effects for simulation-based shifting of average frequency: fricatives were not altered with CI-simulated feedback at all, and vowel quality alterations bore no relationship to the specifics of the spectral manipulation with respect to each speaker’s particular F1/F2 vowel space (Casserly, 2015). Instead of compensation for relative frequency shifts, speakers significantly reduced phonological contrast in F1 while maintaining consistent F2 production.…”

Auditory Learning Using a Portable Real-Time Vocoder: Preliminary Findings

“…A database containing sets of 25 words spoken with either declarative (falling f0) or interrogative (initial fall + final rise in f0) prosodic contours (Casserly, 2013) was used to provide the stimuli for this task. Each word contained only sonorant segments (vowels, glides, liquids and nasals), allowing the f0 information in the acoustic signal to be uninterrupted.…”