Evidence for Relationship Between Speech Perception in Noise and Spectrotemporal Modulation Processing in Children...

Children with mild bilateral hearing loss (CHL), defined as having thresholds between 20 and 40 dB HL, experience difficulties conversing in adverse acoustic conditions with background noise, despite having near-normal speech perception in quiet. Mixed evidence of hearing aid benefit among CHL has led to guidelines that emphasize individualized management, making it challenging for clinicians to suggest intervention options. The overarching goal of this study was to investigate the influence of suprathreshold neural processing on speech perception in noise and how it is influenced by amplification. We focused primarily on auditory temporal and spectral processing, the two main predictors of speech perception in noise among adults with hearing loss.

In 11 neurotypical CHL (6- to 14-yr-old) and 28 age-matched neurotypical children with normal hearing (CNH), the signal to noise ratio required to achieve 70% accuracy on a speech perception in noise task was tracked using a spatial release from masking paradigm. Across three conditions, target words were presented from a front loudspeaker (0°), while noise (2-talker babble) was presented either from the same front loudspeaker or from the left or right loudspeaker (±45° azimuth), with a copy of the noise being presented from the front loudspeaker 3-msec later from the lead noise to simulate spatial separation between the target words and noise. Neural sensitivity to spectral and temporal modulations was assessed objectively using functional near-infrared spectroscopy (fNIRS). The fNIRS responses were recorded in the bilateral frontotemporal area, when children listened to broadband noise that was unprocessed (reference), amplitude-modulated at 4 Hz, or spectrally modulated with spectral ripple noise.

CNH and CHL, whether unaided or aided, benefited similarly from the spatial separation between noise and target speech. Aiding neither enhanced nor deteriorated the benefit of spatial release from masking. fNIRS measures in aided CHL showed increased cortical activation to the temporal and spectral modulations in the temporal cortex (TC) compared to both unaided conditions and CNH. Higher aided than unaided cortical activity in the TC likely reflects enhanced sensitivity to speech-relevant acoustic features with amplification. Further, in CHL, lower cortical activity to temporal modulation in the right frontotemporal cortex and higher activity to spectral modulation in the left TC were associated with lower signal to noise ratio required to achieve 70% accuracy. The associations likely reflect region-specific and cue-specific contributions to speech perception in noise among CHL. No associations were found in CNH.

Our study highlights the value of hearing aid fitting in improving speech-in-noise outcomes in CHL. While behavioral benefits may be subtle or inconsistent, cortical data revealed enhancement in auditory processing with amplification, particularly in the TC. Our findings suggest that suprathreshold neural processing underlies the variation in speech perception in noise among CHL.