14th Speech in Noise Workshop, 12-13 January 2023, Split, Croatia 14th Speech in Noise Workshop, 12-13 January 2023, Split, Croatia

P13Session 1 (Thursday 12 January 2023, 15:30-17:30)
Reaction times to spectrotemporal modulations as a predictor of speech in noise after changes in cochlear-implant device settings

Elisabeth Noordanus
Donders Centre for Neuroscience, Biophysics, Radboud University, Netherlands

Josef Chalupper
Advanced Bionics European Research Center, Hannover, Germany

Marc M. van Wanrooij
Donders Centre for Neuroscience, Biophysics, Radboud University, Netherlands

Lucas H.M. Mens
Radboudumc, Nijmegen, Netherlands

A. John van Opstal
Donders Centre for Neuroscience, Biophysics, Radboud University, Netherlands

Speech perception is a complex cognitive skill that not only depends on the integrity of the (restored) auditory system, but also on factors like language skills, short-time memory, and other cognitive characteristics of the listener. Therefore, speech-perception measures may not be an optimal method to evaluate the effect of clinical interventions, such as changing cochlear implant (CI) fittings. The sensitivity to spectrotemporal (ST) modulations is essential for speech intelligibility, especially temporal modulations between 4 - 32 Hz and spectral modulations up to 1 cycle/kHz are important. We explored reaction time to a modulation onset as an objective measure of the sensitivity to different ST modulations. Reaction times in response to ST modulation onsets may be less contaminated by cognitive stages of auditory processing and therefore suited as early indicators of the effect of changes in device settings. The reaction-time paradigm does not depend on working memory like alternative forced-choice paradigms. Thirty seconds suffice to measure the sensitivity to a certain ST modulation.

We measured the reaction time to an unpredictable onset of a ST modulation embedded in noise for 21 unilateral CI users. We also compared results collected in a supervised and an unsupervised condition. In the supervised condition, sounds were presented in the free field and reaction times were recorded by a spacebar press on a laptop while monitored by the test leader. The unsupervised method employed a research app on a smartphone, using wireless streaming directly to the CI sound processor, while subjects responded by pressing the touchscreen. The reaction-time task was feasible for all subjects. Speech understanding in noise was also assessed in both conditions. The reaction times and the speech in noise results were largely comparable for both methods. We observed strong idiosyncratic differences in the reaction times to the different spectral and temporal modulations onsets. Across subjects, reaction times to ST modulations representative of modulations in speech were correlated with speech in noise performance (r = 0.6 - 0.7). The highest correlations were found for the broadband stimuli with density 0.25 cycles/octave (r = 0.72 for the pure spectral stimulus, 0.66 for 16 Hz), and for 8 Hz 0.5 c/o (r = 0.70). In addition, the 1-octave 4 kHz centered 4 Hz 2 c/o had a good correlation (0.64). In conclusion, reaction times to ST modulations, even in an unsupervised condition, provide an objective measure for the neural sensitivity to ST modulations of CI users and promise to provide an early indicator of the development of speech in noise understanding after device changes.

Last modified 2023-01-06 23:41:06