Flaws in Spectral Ripple Stimuli for Listeners with Cochlear Implants

Abstract

ObjectivesSpectral resolution is an important aspect of hearing speech accurately, and a known limiting factor in cochlear implants (CIs). The spectral ripple discrimination task is a psychophysical measure that has been found to correlate with word recognition, music perception, and various aspects of CI processing. It was hypothesized that the limited spectral sampling of CI processors would distort the output of spectral ripple stimuli in ways that could be problematic for their use in CI research. DesignIdealized acoustic ripple spectra were analyzed through synthetic filters matched to CI processor channels, and the output was examined for spectral density and modulation depth cross a wide range of inputs. Existing data showing correlations between ripple discrimination and speech recognition were re-analyzed with special consideration of scores above the spectral aliasing limit. ResultsThe analysis revealed numerous complicating factors in the transmission of spectral ripples through CIs, including spectral aliasing and modulation depth neutralization. These unintended distortions are nonlinear in nature, with the consequence that spectral ripple thresholds above a certain number cannot be ordered monotonically along a single parametric axis/dimension. For the Cochlear device, the critical limit for spectral aliasing is around 2.1 RPO, and the critical limit for modulation saturation is about 3.5 RPO. Neither limitation is remediated by changing stimulus phase. Changes in modulation depth are limited by the single channel with the narrowest input filter octave-scaled bandwidth (channel 9 in the Cochlear device). Correlations between ripple discrimination and various measures of speech recognition are improved when high ripple scores are dismissed or winsorized. An additional difficulty with interpreting discrimination thresholds is that ripple stimuli often do not match the spectral characteristics of modulations found in actual speech sounds.ConclusionsThe authors conclude that the practice of identifying the highest spectral density at which ripple phase inversion can be detected should be abandoned because (1) it lacks the well-behaved mathematical properties a psychophysical experimenter requires, and (2) high spectral densities are a poor representation of ecologically-relevant cues. Alternative testing strategies (ripple depth detection, ripple phase shift, acoustic-phonetic cue perception) should be taken, with the specific consideration of real speech acoustics in their development.

Spectral aliasing in an acoustic spectral ripple discrimination task

“…Similarly, experiments using SMR in CI users are limited to SMR densities below the limit where stimulus generation introduces aliasing, suggesting that findings in this population, notably the consistent correlation with speech outcomes, were likely robust to this effect (DiNino and Arenberg, 2018;Holden et al, 2016;Lawler et al, 2017). That said, CI digital signal processing requires filtering signals into discrete bands for representation on individual electrodes, a process that under-samples dense SMR stimuli and introduces another source of spectral aliasing (Winn and O'Brien, 2019).…”

InterlACE Sound Coding for Unilateral and Bilateral Cochlear Implants

“…In addition, high ripple numbers lead to aliasing effects [52]- [54]. Due to the underlying filter bank of the CI processing, the ripple resolution is limited by aliasing occurring in the output stimulation sequence with higher ripple numbers starting.…”

Spectrotemporal Modulation Sensitivity in Cochlear-Implant and Normal-Hearing Listeners: Is the Performance Driven by Temporal or Spectral Modulation Sensitivity?

“…The conclusions were the same as those drawn from the Pearson's correlations: The spectrotemporal modulation thresholds were significantly correlated with MDTs (q ¼ À0.67, p ¼ .003) but not with static ripple thresholds (q ¼ 0.21, p ¼ .46). Assuming that the thresholds above the limit do represent better performance but the magnitude cannot be ordered (8 RPO not necessarily better than 7 RPO), the spectrotemporal modulation and static ripple thresholds that were greater than 2.5 RPO were replaced with the value of 2.5 and the analysis was repeated (Winn & O'Brien, 2019). Results remained the same: spectrotemporal modulation thresholds were correlated with MDTs (r ¼ À.59, p ¼ .009) but not with the static ripple thresholds (r ¼ .28, p ¼ .34).…”