Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 May 1.
Published in final edited form as: Ear Hear. 2021 Nov 1;43(3):849–861. doi: 10.1097/AUD.0000000000001144

Effect of noise reduction on cortical speech-in-noise processing and its variance due to individual noise tolerance

Subong Kim a, Yu-Hsiang Wu b, Hari M Bharadwaj a,c, Inyong Choi b,d,*
PMCID: PMC9010348  NIHMSID: NIHMS1739420  PMID: 34751679

Abstract

Objectives:

Despite the widespread use of noise reduction (NR) in modern digital hearing aids, our neurophysiological understanding of how NR affects speech-in-noise perception and why its effect is variable is limited. The current study aimed to 1) characterize the effect of NR on the neural processing of target speech and 2) seek neural determinants of individual differences in the NR effect on speech-in-noise performance, hypothesizing that an individual’s own capability to inhibit background noise would inversely predict NR benefits in speech-in-noise perception.

Design:

Thirty-six adult listeners with normal hearing participated in the study. Behavioral and electroencephalographic responses were simultaneously obtained during a speech-in-noise task in which natural monosyllabic words were presented at three different signal-to-noise ratios, each with NR off and on. A within-subject analysis assessed the effect of NR on cortical evoked responses to target speech in the temporal-frontal speech and language brain regions, including supramarginal gyrus (SMG) and inferior frontal gyrus (IFG) in the left hemisphere. In addition, an across-subject analysis related an individual’s tolerance to noise, measured as the amplitude ratio of auditory-cortical responses to target speech and background noise, to their speech-in-noise performance.

Results:

At the group level, in the poorest signal-to-noise ratio condition, NR significantly increased early SMG activity and decreased late IFG activity, indicating a switch to more immediate lexical access and less effortful cognitive processing, although no improvement in behavioral performance was found. The across-subject analysis revealed that the cortical index of individual noise tolerance significantly correlated with NR-driven changes in speech-in-noise performance.

Conclusions:

NR can facilitate speech-in-noise processing despite no improvement in behavioral performance. Findings from the current study also indicate that people with lower noise tolerance are more likely to get more benefits from NR. Overall, results suggest that future research should take a mechanistic approach to NR outcomes and individual noise tolerance.

1. Introduction

Hearing aid (HA) users often have difficulty understanding speech in noisy listening environments, even with prescriptive amplification. Thus, most digital HAs implement signal processing strategies designed to attenuate background noise (Plomp, 1994; Takahashi et al., 2007). The current study focuses on a widely used noise reduction (NR) algorithm that counteracts noise by reducing the gain from noise-contaminated frequency bands (Dillon & Lovegrove, 1993; Hagerman & Olofsson, 2004; Levitt, 2001). However, the speech signal and noise often overlap in frequency content resulting in some inevitable distortions of speech cues while attenuating noise with NR (e.g., Kates (2008)). Although NR may not help improving speech intelligibility mainly due to the tradeoff between noise attenuation and speech cue distortions (e.g., Bentler et al. (2008); Dillon and Lovegrove (1993); Jamieson et al. (1995)), it is apparent from laboratory research, to date, that NR can provide some benefits to listeners’ perception and behavior, including improved ease and comfort of listening and reduced cognitive load (or listening effort) (e.g., Bentler et al. (2008); Mueller et al. (2006); Ohlenforst et al. (2017); Ricketts and Hornsby (2005); Sarampalis et al. (2009); Wendt et al. (2017)).

Previous studies have sought to assess potential NR benefits using a series of outcome measures. Self-reported ratings (e.g., Humes (1999) and Hornsby (2013)) are perhaps the most feasible measures to evaluate subjective NR benefits. Objective measures such as dual-task paradigms have also been notably used to study the cognitive benefits from NR, reflected in improved performance on a secondary task (e.g., Desjardins and Doherty (2014); Sarampalis et al. (2009)). However, these measures cannot provide online estimates of listeners’ cognitive processing difficulty during speech-in-noise tasks. Using physiological measures such as pupillometry or electroencephalography (EEG), researchers have tried to identify NR-driven systematic changes in pupil responses (Ohlenforst et al., 2018; Wendt et al., 2017) or the auditory N1 component (Bernarding et al., 2013; Bernarding et al., 2017; Strauss et al., 2010) that may represent speech processing effort during the listening tasks. Yet, we do not fully understand the neural substrates of potential NR benefits that would occur during the course of speech recognition in noise. More specifically, it is not clear how NR affects online cortical processing involved while listening to noisy speech in terms of the loci and timing of cortical evoked responses.

Prevailing models of lexical processing (Gow, 2012; Hickok & Poeppel, 2007) suggest that spoken words are processed through two distinct higher-cortical pathways: the ventral and dorsal streams. These two streams operate in parallel and take different roles. The ventral stream mediates the sound-to-meaning mapping across the posterior middle temporally gyrus (MTG) and adjacent regions (Davis & Johnsrude, 2003). In contrast, the left-dominant dorsal stream mediates the mapping sound to articulation across the supramarginal gyrus (SMG) and parietal operculum (Rauschecker & Scott, 2009). Both cortical pathways diverge from the inferior parietal lobule and superior planum temporale and then end together at the inferior frontal gyrus (IFG) (Friederici, 2015). The literature highlights SMG, MTG, and IFG in the left hemisphere that show more robust responses to natural words than pseudo-speech sounds (see review, Davis (2016); Gow (2012)). Notably, the dorsal stream pathway consisting of the left SMG and left IFG is known to be related to phonological discrimination rather than semantic processing (via the ventral stream, including MTG) (Caplan et al., 1995; Gow & Caplan, 1996). Functional magnetic resonance imaging (MRI) studies support this idea, reporting that changes in noise level alter the level of blood-oxygen-level-dependent responses to speech across these regions (Du et al., 2014, 2016; Wild et al., 2012; Wong et al., 2009). In terms of timing of speech processing, previous eye-tracker studies using the Visual World Paradigm have explored the time-course of spoken-word recognition. As the spoken-word unfolds over time, dynamic lexical competition occurs naturally and automatically in ideal listening conditions (Farris-Trimble & McMurray, 2013; Huettig & Altmann, 2005). However, with the degraded representation of speech in challenging listening conditions (e.g., high-level noise), this competition could be delayed by up to 250 ms (Ben-David et al., 2011; Farris-Trimble et al., 2014; McMurray et al., 2017). Yet, the time-course of cortical speech processing has not been extensively studied at the higher-cortical regions (i.e., SMG and IFG) beyond the auditory cortex, notably with respect to HA signal processing schemes such as NR.

Emerging evidence also indicates that HA users react very differently to the tradeoff between attenuated noise and distorted speech cues (Neher, 2014; Neher et al., 2014; Neher et al., 2016). This may arise from differences in tolerance to background noise and speech cue distortions (Neher & Wagener, 2016); some listeners could be more bothered by noise, whereas others could be more affected by speech cue distortions induced by NR. In addition to hearing sensitivity and cognitive performance that partially explains inter-individual variability in NR outcomes and preference (Neher, 2014; Neher et al., 2014; Neher et al., 2016), researchers have tried to explore what factors drive differences in individual tolerance to noise and speech cue distortion using a range of measures, including psychoacoustic, audiological measures, and self-reported ratings (e.g., Brons et al. (2014); Neher and Wagener (2016)). However, none of those measures could significantly capture inherent factors that contribute to individual tolerance. Previous studies also combined physiological and behavioral measures to understand the biological mechanisms underlying the variability in speech-in-noise perception by exploring the relationship between behavior and electrophysiology using cortical auditory evoked potentials (Billings et al., 2013; Billings et al., 2015; Parbery-Clark et al., 2011) and neural envelope tracking (Decruy et al., 2020; Vanthornhout et al., 2018). However, little attention has been directed to developing cortical measures of an individual’s ability to inhibit background noise and understanding the relationship between individual noise tolerance and perceptual benefits from NR.

In our recent study on normal-hearing listeners (Kim et al., 2021), we investigated how the change in noise level, without additional signal processing (e.g., NR) artifact, affects cortical speech-in-noise processing over time across the dorsal stream pathway. Our source-space analysis revealed that reduced noise significantly led to an increase in SMG activity at around 300 ms after word onset during speech-in-noise tasks in which, on average, the target word (which was monosyllabic natural word such as ”pan”) ends at 500 ms, indicating that the reduction of the noise level elicited efficient and fast speech processing. In terms of individual differences in speech-in-noise outcomes, Kim et al. (2021) quantified individual noise tolerance by calculating the neural signal-to-noise ratio (SNR), the amplitude ratio of cortical evoked responses to target speech relative to noise at the auditory cortex during speech-in-noise tasks, similar to the “neural” form of a target-to-masker ratio (Hillyard et al., 1998; Mesgarani & Chang, 2012). We found that listeners show large variance in this cortical index of noise tolerance and that listeners with higher neural SNR have greater speech-in-noise performance (i.e., accuracy).

In the present study, first, we aimed to characterize how well NR could facilitate cortical speech-in-noise processing. As a follow-up to the study by Kim et al. (2021), we used the same regions-of-interest (ROIs) (i.e., left SMG and left IFG) to assess the effect of NR on cortical evoked responses across the dorsal lexicon areas at different noise levels in a within-subjects design. It should be noted that, in addition to the noise attenuation effect, NR induced undesirable speech cue distortions that might lead to greater processing effort (Ward et al., 2017; Winn et al., 2015). We hypothesized that despite distorted speech cues, attenuated noise from NR still likely to result in more efficient and fast speech-in-noise processing reflected in increased SMG and decreased IFG activity, especially at the most challenging SNR condition. This hypothesis was based on previous findings showing 1) reduced cognitive effort due to NR only in their most challenging listening condition (e.g., Desjardins and Doherty (2014); Sarampalis et al. (2009)) and 2) that changes in noise level have opposite effects on SMG and IFG activity (Kim et al., 2021).

Secondly, in terms of individual differences in NR outcomes, we explored a biology-guided approach to individual outcomes. As in Kim et al. (2021), we quantified each individual’s noise tolerance by measuring neural SNR and identified whether the neural SNR could predict NR outcomes in an across-subject design. We expected that people with lower noise tolerance (i.e., lower neural SNR) would appreciate NR more, showing an increase in behavioral accuracy compared with no NR condition.

2. Materials and Methods

2.1. Participants

Thirty-six normal-hearing native speakers of American English between 18 and 34 years of age (mean = 22.72 years, SD = 3.51 years, 6 (17%) male) took part in our experiments. All participants were recruited from a population of students at the University of Iowa. Pure-tone hearing thresholds were measured using a GSI-61 audiometer (Grason-Stadler Inc., Littleton, MA) through TDH-39P circum-aural headphones (Telephonics Corporation, Farmingdale, NY) at octave frequencies from 250 to 8000 Hz as well as at 3, 6, and 12 kHz (Figure 1). The pure-tone thresholds at 0.5, 1, 2, and 4 kHz no greater than 20 dB HL, as well as hearing symmetry within 20 dB for any tested frequency, were used as our hearing screening criteria. All study procedures were reviewed and approved by the University of Iowa Institutional Review Board. All work was carried out in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki), and written informed consent was obtained for everyone.

Figure 1.

Figure 1.

Open in a new tab

Average pure-tone hearing thresholds for participants. The error bars indicate ±1 standard deviation.

2.2. Task Design and Procedures

The experimental task was a speech-in-noise recognition test in which we simultaneously measured behavioral performance and cortical evoked responses. We used consonant-vowel-consonant monosyllabic English words (Geller et al., 2020), spoken by a male and a female speaker with a general American accent, embedded in speech-shaped stationary noise. Participants were tested in a single-walled, sound-treated booth, and the stimuli were presented to both ears through insert ER-2 earphones (Etymotic Research, Elk Grove, IL) with ER1–14A/B ear tips (Etymotic Research, Elk Grove, IL). A computer monitor was placed at 0-degrees azimuth 0.5 m distance from the participant at eye level. The task was implemented in MATLAB (R2016b, the MathWorks, Natick, MA) using the Psychtoolbox-3 package (Brainard, 1997; Pelli, 1997).

Every trial in the task (Figure 2) started with presenting a trial number in silence on the screen for a half-second. Then, a fixation cross (‘+’) in silence was presented at the center of the screen and remained throughout the trial. In order to assure minimizing eye-movement artifacts, participants were instructed to fix their gaze on this symbol throughout the stimulus presentation. After 0.5 seconds of silence with the fixation cross, the speech-shaped stationary noise began and continued for 1.5 seconds, and 0.5 seconds after the noise onset, the target word was presented. The composite stimulus (noise and word) was normalized for root-mean-square (RMS) amplitude to 65 dBA. After the auditory stimulus was presented, participants chose one answer from the list of four-word options on the screen (i.e., a four-alternative forced-choice test), which differed in an initial consonant (e.g., sat, pat, fat, that for the target word sat), using a keypad. No feedback was provided to the participant. As in Kim et al. (2021), we designed a noise-only period before presenting the target word to derive distinct auditory-cortical responses to the onsets of noise and the target word, respectively, and calculate the amplitude ratio between those evoked responses to index individual noise tolerance. Words were presented at three input SNR conditions set for each participant.

Figure 2.

Figure 2.

Open in a new tab

Trial and stimulus structure. In this example, the target word “sat” was presented.

Using an adaptive test with the two-down-one-up procedure with a fixed level of speech (Bernstein & Gravel, 1990), the first 50 trials examined the SNR level in which 70% of words were recognized correctly (i.e., speech reception threshold (SRT-70)) for each participant. We here named the SRT-70 the medium input SNR that varied from −8 to −3 dB (median: −5.5 dB) across individuals. A participant’s high and low input SNR conditions were determined based on the individual’s SRT-70 (i.e., the SNRs 3-dB higher or lower than SRT-70). The whole experimental session comprised six different conditions: three SNR conditions, each with NR off and on. After SRT-70 was measured, the next 330 trials (six sets of 55 words balanced across speaker gender and initial phonemes) were randomly assigned to each experimental condition.

2.3. NR Algorithm

The present study used the Ephraim-Malah NR algorithm (Ephraim & Malah, 1984, 1985), which is one of the NR algorithms implemented in modern digital HAs (e.g., Pisa et al. (2010); Sarampalis et al. (2009); Stelmachowicz et al. (2010)). This modified spectral-subtraction NR attenuates more gain across frequencies that are more contaminated by noise based on SNR estimation in each short-time spectrum (For more details about the NR algorithm, see Ephraim and Malah (1984, 1985)). Figure 3A illustrates how fast the NR algorithm used in the current work was engaged, reacting to SNR estimation in each short-time frame (Figure 3A only shows the estimated SNR and applied noise attenuation averaged across frequencies at each time-point). Once NR was fully activated at around 200 ms after the noise onset (i.e., before the target word was heard), noise attenuation was applied almost instantaneously according to changes in estimated SNR.

Figure 3.

Figure 3.

Open in a new tab

A. The implementation of the noise reduction (NR) algorithm. B. Power spectral density of speech and noise stimuli extracted using the phase-inversion technique. C. Magnitude-squared coherence across frequencies for two speech stimuli (NR off vs. on). SNR: signal-to-noise ratio.

Different spectral gain reductions could be provided for the same hearing loss configuration depending on NR algorithms, even when using the same stimuli (Hoetink et al., 2009). For this reason, accurate quantification of SNR changes and speech cue distortions due to NR is vital to validate the NR outcomes (Gustafson et al., 2014). Thus, we quantified SNR changes using the phase-inversion technique (Hagerman & Olofsson, 2004) that has been implemented in HA evaluations (e.g., Hagerman and Olofsson (2004); Naylor and Johannesson (2009); Wu and Bentler (2007, 2009, 2011); Wu and Stangl (2013)). This technique processes two noisy signals that are identical except having a different phase in noise and then adds the two outputs or deduct one from the other to extract signal and noise stimuli, respectively. Using the extracted speech and noise stimuli, we could calculate the SNR (Figure 3B). In this work, the NR algorithm provided an approximately 3.5-dB benefit in SNR based on long-term RMS level, consistent across different input SNRs and different sets of target words. In terms of quantifying speech cue distortions, we measured magnitude-squared coherence that has also been widely used in HA studies (e.g., Kates and Arehart (2005); Lewis et al. (2010); Ma and Loizou (2011)). Coherence represents the relationship between input and output power spectrums, ranging from zero (i.e., totally different) to one (i.e., identical). In the present study, we measured the coherence across frequencies between unprocessed and NR-processed speech stimuli, extracted using the inversion method, and found a greater reduction in coherence across the frequency range in which greater level differences occurred on power spectral density for the speech stimuli (Figure 3C). The average coherence up to 12 kHz was at around 0.6 and consistent across different input SNRs and different sets of target words.

2.4. Data Acquisition and Pre-processing

EEG data were recorded during the speech-in-noise task using the BioSemi ActiveTwo 64-channel system (BioSemi B.V., Amsterdam, the Netherlands) at a 2048 Hz sampling rate in the international 10–20 layout. Data from each electrode were bandpass filtered from 1 to 50 Hz using a 2048-point zero-phase finite-impulse-response filter and re-referenced to the average across two mastoid channels. Then, epochs were extracted from −500 ms to 2 s relative to the noise onset, and baseline-corrected using the average amplitude between −200 and 0 ms. Each epoch was down-sampled to 256 Hz. After removing ocular, muscular, and cardiac artifacts using independent component analysis (Jung et al., 2000), the average across trials in each experimental condition was calculated for each scalp electrode to make evoked event-related potentials (ERPs). We averaged across all the trials within each condition irrespective of response accuracy because we aimed to observe neural processes that happen “prior” to the subjects’ behavioral responses. Also, we wanted to avoid omitting some word items during averaging in an uncontrolled way since it could affect the ERP morphology.

2.5. Source-space Analysis

In order to understand underlying neural substrates of NR effects and individual noise tolerance, source-level EEG analysis was considered given that electrode data would not locate an underlying source (Makeig et al., 1996) due to several resistive layers of the scalp and the skull (Nunez & Westdorp, 1994; Srinivasan et al., 1996). In the present study, the distribution of source activity was calculated based on minimum norm estimation (Gramfort et al., 2013; Gramfort et al., 2014), assuming multiple sparse priors (Friston et al., 2008). Although we did not have individual structural MRI data, we used a locator device (PATRIOT, Polhemus, Colchester, VT) to co-register the position of electrode channels obtained from each participant to the average MRI. Then, we calculated the forward solution using the FreeSurfer’s “fsaverage” source model and boundary-element-method head model (Fischl et al., 1999; Hämäläinen & Sarvas, 1989), and noise covariance across channels from each participant. Finally, the inverse operator was calculated using the forward solution and the estimation of noise covariance from each individual and applied to epochs using a regularization parameter of 1/32 and dynamic statistical parametric maps (dSPMs) as a noise-normalization procedure (Dale et al., 2000), to estimate source-space ERP time-courses across 4098 cortical vertices per hemisphere.

In this work, our ROIs were defined based on the literature review. We explored left SMG and left IFG (see review, Davis (2016); Gow (2012)) for the within-subject analysis of NR/SNR effects on evoked source activity, and right Heschl’s gyrus (i.e., the primary auditory cortex) (e.g., Alexander et al. (1996); Carcea et al. (2017); Gilmore et al. (2009); Kim et al. (2021); Mesgarani and Chang (2012)) for the across-subject analysis regarding individual differences in NR outcomes.

Due to multiple vertices included in each of the ROIs, the whole-brain analysis requires strict control of multiple comparisons, which could significantly reduce statistical power. There are many alternative analytical approaches, including calculating the mean or median across vertices in the region. However, since we have no individual MRI head models, taking the summed activity might lead to spatial blurring due to individual variability in the brain’s functional and anatomical structure. Instead, we identified a representative vertex for each region-of-interest in each experimental condition by calculating the cross-correlation coefficients across vertices in the brain region over time-points and then choosing a vertex with the maximum value of mean coefficients (Tong et al., 2016).

For both within- and across-subject analyses, temporal ERP envelopes were extracted from source cortical activity, as we did not rely on a single ERP component. This was done by applying the Hilbert transform to 3–8 Hz bandpass-filtered data and then calculating the absolute value. We used this frequency range capturing neural oscillations from both early temporal and late frontal lobes (Giraud & Poeppel, 2012) and chose a zero-phase finite-impulse-response filter whose impulse response was symmetric and non-causal (de Cheveigne & Nelken, 2019).

2.6. Statistical Analysis

For the within-subject analysis, first, a two-way repeated-measures analysis of variance (ANOVA) was performed on the influence of SNR and NR on behavioral accuracy to investigate 1) if SNR manipulation was sufficient to secure different levels of difficulty in speech perception between listening conditions and 2) if the present study replicated previous findings showing no positive effect of NR on behavioral accuracy (e.g., Bentler et al. (2008); Dillon and Lovegrove (1993); Jamieson et al. (1995)). Then, we conducted a two-way repeated-measures ANOVA on the effect of NR and SNR on cortical activity. Because we had a priori hypothesis that the NR benefits in cortical speech processing would be different across SNR conditions based on previous findings (e.g., Desjardins and Doherty (2014); Sarampalis et al. (2009)), we aimed to examine the NR effect in each SNR. Therefore, post hoc paired t-tests were performed with the Bonferroni correction, regardless of main effects and interactions (Hsu, 1996; Maxwell & Delaney, 2004). In addition, to investigate whether the NR effects on cortical activity were opposite between SMG and IFG, in line with our a priori hypothesis based on Kim et al. (2021), 2 × 2 repeated measures ANOVAs with ROI and SNR as the within-subject factors were performed in each SNR condition. For neural data, we calculated a range of leave-one-out grand averages (i.e., jackknife approach) before testing to conduct statistical analysis with making use of clean ERPs from each grand average. Then, we examined the effect of NR on SMG and IFG activity in multiple input SNRs, using the peak magnitude of temporal envelopes obtained over different time-points: 100–400 ms for SMG and 400–700 ms after the word onset for IFG. Inflated F-value and t-value due to the jackknife approach that artificially reduced the error variance were adjusted by dividing the resulting F-value and t-value by (N - 1)2 and N – 1, respectively (Luck, 2014).

For the across-subject analysis, Pearson correlation analysis was conducted to measure a linear association between individual noise tolerance indexed by the neural SNR and NR benefits (i.e., NR on vs. off) in behavioral accuracy. As in Kim et al. (2021), the neural SNR was expected to estimate an individual’s noise tolerance, which was computed into a dB scale using the amplitude ratio of auditory-cortical responses to the target word and noise during the speech-in-noise task. For the computation of the neural SNR, we used the peak magnitude of temporal envelopes extracted from ERPs averaged across three NR-off SNR conditions, within a 100–400 and 200–500 ms time-window following the noise and word onset, respectively. Changes in behavioral accuracy (%) were also averaged across three SNR conditions. A post hoc two-sample t-test where degrees of freedom were estimated using the Satterthwaite approximation (Satterthwaite, 1946) was performed to compare the neural SNR between two groups of participants divided based on their behavioral accuracy; the group of listeners who showed improvement (i.e., changes in performance greater than zero) due to NR and the other group with the opposite case were compared regarding their inherent noise tolerance reflected in the neural SNR.

3. Results

3.1. Behavioral Performance

Behavioral performance (accuracy) varied among participants. This was observed in the high SNR condition (NR off: mean = 83.59%, SD = 6.41%; NR on: mean = 81.72%, SD = 6.90%), the medium SNR condition (NR off: mean = 75.35%, SD = 7.76%; NR on: mean = 72.73%, SD = 8.33%), and the low SNR condition (NR off: mean = 62.42%, SD = 10.07%; NR on: mean = 59.60%, SD = 10.96%) where the chance level was at 25%.

To determine the effect of NR on behavioral accuracy in multiple input SNRs, a two-way repeated-measures ANOVA was performed. The within-subjects independent variables were SNR (high vs. medium vs. low) and NR status (off vs. on), and the dependent variable was speech-in-noise performance (Figure 4). The statistical tests were performed on the rationalized arcsine units (Studebaker, 1985). Significant main effects of SNR (F2,70 = 170.83, p < 0.001) and NR (F1,35 = 6.31, p = 0.017) were observed. The interaction between input SNR and NR was not significant (F2,70 = 0.022, p = 0.98). The post hoc comparison between SNR conditions revealed significantly different performance between each pair (high vs. medium SNR: t(35) = 7.96, adjusted p < 0.001; high vs. low SNR: t(35) = 18.43, adjusted p < 0.001; medium vs. low SNR: t(35) = 10.47, adjusted p < 0.001). These results indicated 1) that the SNR manipulation in the current work was adequate to make significant differences in difficulty between input SNR conditions and 2) a slight but significant performance decrement with NR as expected from previous findings in adults showing either no improvement with NR (e.g., Alcántara et al. (2003); Bentler et al. (2008); Ricketts and Hornsby (2005)) or negative effect on the performance (e.g., Jamieson et al. (1995); Kates (2008)).

Figure 4.

Figure 4.

Open in a new tab

Box plot of behavioral accuracy across each condition. The edges of the box represent the 25th and 75th percentile. The central bars on each box mark the median. Solid lines denote changes in performance between conditions for the same subject. SNR: signal-to-noise ratio, NR: noise reduction, ***: significant at p < 0.001, *: significant at p < 0.05, n.s.: not significant.

3.2. Auditory Evoked Potentials at the Sensor Space

Before conducting source-space analysis, the quality of EEG data was checked at the sensor-space level using grand-average evoked potentials averaged across the frontal-central channels (C1, C2, FC1, FC2, FCz, and Cz). Figure 5 show clear auditory components (e.g., N1, P2) elicited at both the noise and target-word onsets from the waveforms in each SNR condition (left panel) and the average across SNRs in NR off and on conditions (right panel). The data quality verified by the sensor-space analysis showing prominent frontal-central ERP components occurring at a typical N1-P2 time window allowed us to examine evoked source activity from the dorsal stream pathway of speech processing (i.e., SMG and IFG) for the within-subject analysis and Heschl’s gyrus for the across-subject analysis.

Figure 5.

Figure 5.

Open in a new tab

Exemplar waveforms for the data-quality check. Grand-average evoked potentials averaged across the frontal-central channels (C1, C2, FC1, FC2, FCz, and Cz, denoted as black dots in the electrode-position legend within the left panel) are shown. The left panel shows the waveforms in each signal-to-noise ratio (SNR), and the right panel shows the averaged waveforms across SNRs in the noise reduction (NR) off and on conditions. Gray waveforms show evoked potentials at other electrodes (i.e., gray dots in the electrode-position legend).

3.3. Effect of NR on Evoked Source Activity in Multiple Input SNRs

To investigate whether the effects of NR on cortical speech processing were different across input SNR conditions, a two-way repeated-measures ANOVA was conducted. The SNR (high vs. medium vs. low) and NR status (off vs. on) were adopted as the within-subjects independent variables, and the peak magnitude of temporal ERP envelopes from left SMG (Figure 6B) and left IFG (Figure 6E), respectively, served as the dependent variable. Those peak source activities were obtained over 100–400 ms for SMG (Figure 6A) and 400–700 ms after the word onset for IFG (Figure 6D), respectively.

Figure 6.

Figure 6.

Open in a new tab

Region-of-interest (ROI) based source-space analysis. A and D. The time-course of grand-average evoked source activity (thin lines) and its temporal envelope (thick lines), with the standard error of the mean (±1 SEM). B and E. Mean peak magnitude of the temporal envelope, with ±1 SEM. C and F. Whole-brain maps captured at the peak-activity timing in the low signal-to-noise ratio (SNR) condition show the test statistics from post hoc paired t-tests in each cortical vertex in the left hemisphere. SMG: supramarginal gyrus, IFG: inferior frontal gyrus, dSPM: dynamic statistical parametric maps, NR: noise reduction, **: significant at p < 0.01, *: significant at p < 0.05, n.s.: not significant.

For SMG activity, significant main effect of NR was observed (F1,35 = 11.21, p = 0.0020), whereas main effect of input SNR was not significant (F2,70 = 0.45, p = 0.64). The interaction between NR and input SNR was not significant (F2,70 = 0.70, p = 0.50). The post hoc analysis (paired t-test) revealed that SMG activity significantly increased due to NR processing only in the low SNR condition (t(35) = −2.64, adjusted p = 0.037, effect size = 0.62). For IFG activity, no significant main effects of NR (F1,35 = 1.060, p = 0.31) and input SNR (F2,70 = 0.92, p = 0.40) were observed. No significant interaction between NR and input SNR was found (F2,70 = 2.26, p = 0.11). Post hoc analysis suggested that NR processing significantly decreased IFG activity in the low SNR condition (t(35) = 2.54, adjusted p = 0.047, effect size = 0.55), although the global effect of NR was not significant. Significant changes in SMG/IFG activity due to NR processing in the low SNR condition were also validated in whole-brain maps obtained at the peak-activity timing, illustrated in Figure 6 C, F.

Figure 7 compares the pattern of NR effects on cortical activity between two ROIs. In the most challenging SNR condition (i.e., low SNR), no significant main effects of NR (F1,35 = 0.22, p = 0.64) and ROI (F1,35 = 0.13, p = 0.72) were observed as NR processing had opposite effects on SMG and IFG activity. However, the interaction between NR and ROI was significant (F1,35 = 10.68, p = 0.0024), indicating that the NR effect on cortical activity significantly differed between the ROIs (Figure 7 rightmost panel). On the contrary, no significant main effects and interaction between NR and ROI were revealed in the high SNR condition (NR (F1,35 = 0.34, p = 0.56), ROI (F1,35 = 0.79, p = 0.38), and interaction (F1,35 = 0.050, p = 0.82)). In the medium SNR, although significant main effects of NR (F1,35 = 4.31, p = 0.045) and ROI (F1,35 = 4.15, p = 0.049) were observed, no significant interaction (F1,35 = 0.94, p = 0.34) was revealed.

Figure 7.

Figure 7.

Open in a new tab

Region-of-interest (ROI) based source-space analysis comparing mean peak magnitude of the temporal envelope across two ROIs, with ±1 standard error of the mean. SMG: supramarginal gyrus, IFG: inferior frontal gyrus, dSPM: dynamic statistical parametric maps, SNR: signal-to-noise ratio, NR: noise reduction, **: significant at p < 0.01, *: significant at p < 0.05, n.s.: not significant.

Noise attenuation via NR increased SMG and decreased IFG activity in the low SNR condition (i.e., 3-dB below SRT-70) where the noise level was the greatest, which was consistent with our a priori hypothesis. These results indicated that NR facilitated speech processing through SMG and made the IFG activity more resistant to the effects of noise despite inevitable speech cue distortions.

3.4. Individual Differences in Noise Tolerance Predict NR Benefits in Behavioral Accuracy

To examine whether individual noise tolerance indexed by the neural SNR correlated with NR benefits (NR on vs. off) in behavioral accuracy, Pearson correlation analysis was performed. The neural SNR, the amplitude ratio of auditory-cortical responses to the target word relative to background noise averaged across three NR-off SNR conditions, showed a significant correlation (r = −0.44, p = 0.0066) with changes in behavioral accuracy due to NR (Figure 8A). For a post hoc two-sample t-test on the neural SNR, participants were divided into two groups based on their behavioral accuracy changes in which the cut-off was zero. One participant group (Group 1 in Figure 8B, N=13) showed improvement (greater than zero) in their performance due to NR (NR on vs. off, mean Δ = 3.73%, SD = 2.14%), whereas the other group of participants (Group 2 in Figure 8B, N=23) showed no improvement or deterioration (smaller than zero) in performance (NR on vs. off, mean Δ = −5.93%, SD = 3.97%). The present study chose a simple yet logical cut-off, and one-sample t-test results showed that the group means were significantly different from zero (Group 1: t(12) = 6.022, p < 0.001; Group 2: t(22) = −7.019, p < 0.001).

Figure 8.

Figure 8.

Open in a new tab

Individual differences in noise tolerance and noise reduction (NR) benefits. A. Individual noise tolerance indexed by the neural signal-to-noise ratio (SNR), the amplitude ratio of auditory-cortical responses to the target word and noise during speech-in-noise tasks, correlates with changes in behavioral accuracy due to NR. B. Illustration of the computation of neural SNR and differences in auditory-evoked responses between two groups divided based on NR-driven changes in behavioral accuracy. The thin lines indicate grand-average auditory-evoked responses, while the thick lines indicate their temporal envelope, with ±1 standard error of the mean at the peak-activity timing. dSPM: dynamic statistical parametric maps.

Individual noise tolerance reflected in the neural SNR was significantly different (t(30.60) = −2.13, p = 0.041) between these two groups. Figure 8B illustrates differences between two participant groups in their auditory-evoked responses to noise onset and target-word onset, indicating inherent differences in noise tolerance between the two groups. These results were consistent with our a priori hypothesis, suggesting that listeners with lower noise tolerance (i.e., lower neural SNR), e.g., Group 1 in Figure 8, were likely to get more NR benefits in speech-in-noise performance, compared to listeners with higher noise tolerance, e.g., Group 2 in Figure 8.

4. Discussion

4.1. Underlying Neural Mechanisms of NR Effects on Speech-in-noise Processing

The current study revealed that as NR attenuated the background noise, listeners showed significantly increased activity in left SMG at around 300 ms and decreased activity in left IFG at around 500 ms after the word onset when the input SNR was the poorest (i.e., low SNR condition). These results indicated that NR could facilitate cortical speech-in-noise processing through SMG and reduce cognitive processing effort by less engaging IFG despite no improvement in behavioral accuracy. Previous eye-tracker studies using the Visual World Paradigm reported that dynamic lexical competition occurring during the course of spoken-word recognition peaked at around 300–400 ms after the onset of the target word (Farris-Trimble & McMurray, 2013; Huettig & Altmann, 2005) while challenging listening conditions could delay this processing by up to 250 ms (Ben-David et al., 2011; Farris-Trimble et al., 2014; McMurray et al., 2017). The time-window of SMG/IFG activity used in the present study well corresponded to the time-course of lexical processing delineated in these eye-tracker studies, indicating that the spatiotemporal pattern of cortical activity across the dorsal-stream pathway could be a neural marker of more immediate lexical access and less effortful cognitive processing that became available for listeners due to NR processing. These results are in line with the study by Kim et al. (2021) showing that lowering noise level, without engaging artificial spectral distortions, leads to increased evoked responses to the target speech at left SMG and reduced responses at left IFG. Note that NR processing in the present study involved mild speech cue distortions and still provided benefits in cortical speech processing.

The question remains: does this spatiotemporal ERP pattern reflect changes in noise level or speech quality? Previous literature suggested that speech degradation due to either high-level noise or distorted speech (e.g., vocoded sounds) could lead to delayed lexical processing reported in eye-tracker studies (Ben-David et al., 2011; Farris-Trimble et al., 2014; McMurray et al., 2017) and changes in blood-oxygen-level-dependent responses in the frontal areas such as left IFG reported in functional MRI studies (Du et al., 2014, 2016; Wild et al., 2012; Wong et al., 2009). However, recall that NR provides conflicting effects of noise attenuation and speech cue distortions (e.g., Kates (2008)). The cortical pattern of efficient speech processing revealed in the present study seemed to reflect the overall benefits provided by NR. If the level of speech cue distortions were severe enough to offset the noise-attenuation effect, listeners would be less likely to receive the benefits from NR due to increased speech-processing effort (Ward et al., 2017; Winn et al., 2015). Using a range of NR strength could help examine in part the relative contribution of the conflicting NR’s effects to cortical speech processing efficacy.

Significant differences in the cortical pattern of speech processing between NR-off and -on conditions were observed only in the low SNR condition. This is consistent with the literature (e.g., Desjardins and Doherty (2014); Sarampalis et al. (2009)) showing that NR benefits are more likely to occur at their most challenging SNR levels. Notably, Sarampalis et al. (2009), who also used the Ephraim-Malah NR algorithm (Ephraim & Malah, 1984, 1985), examined the effect of NR on cognitive effort using the dual-task paradigms. Their results suggested that cognitive benefits from NR processing—reflected in better word-recall performance and quicker reaction times—were revealed only in the lowest SNR condition despite no improvement in speech intelligibility.

4.2. Neural Marker of Individual Noise Tolerance That Determines NR Benefits

The present study showed that individual noise tolerance, indexed by calculating the neural SNR of auditory-cortical responses to the target speech relative to noise in the NR-off conditions, could predict NR benefits in behavioral accuracy. This finding is in line with Kim et al. (2021), in which a significant association between the neural SNR and behavioral accuracy (without NR) was found.

The variance in the neural SNR may stem from a combination of differences in peripheral sensory processing and top-down efficacy for target-speech unmasking, as suggested by previous literature (Hillyard et al., 1998; Mesgarani & Chang, 2012) showing the increased “neural” target-to-masker ratio resulted from successful extraction of target sounds from background noise (i.e., auditory scene analysis (Bregman, 1990)). The neural process of auditory scene analysis includes sensory encoding of spectro-temporal features of sounds (Liberman et al., 2016), auditory object formation by grouping those acoustic features (Darwin, 1997), and selectively attending to target objects among competing background noise (Choi et al., 2013; Hillyard et al., 1973; Woldorff, 1993). Evidence indicated that each of these neural processes varies across individuals (Bharadwaj et al., 2015; Bharadwaj et al., 2014; Bressler et al., 2017; Choi et al., 2014; Ruggles et al., 2011; Teki et al., 2016; Teki et al., 2013). Differences between Group 1 and Group 2 in Figure 8B regarding cortical auditory responses to target words and noise may correspond to the concept of sensory gain control (Hillyard et al., 1998) or neural filtering (Obleser & Erb, 2020). Although the neural SNR in the current work only quantified individual noise tolerance rather than disentangled potential contributors from bottom-up and top-down neural processing, the variance in the neural SNR was likely to be driven by differences in auditory nerve integrity (Bharadwaj et al., 2019; Mai et al., 2019) and selective auditory attention (Bressler et al., 2017; Choi et al., 2014). Notably, previous findings reported that selective attention efficacy varies in normal-hearing listeners and modulates cortical auditory responses (Choi et al., 2014; Dai & Shinn-Cunningham, 2016; O’Sullivan et al., 2019) or speech-EEG coherence (Viswanathan et al., 2019). These results indicated that inter-individual variance in noise tolerance due to differences in the ability to unmask target speech from noise could determine the NR benefit in behavioral accuracy for a given listener.

Alternatively, individual differences in NR outcomes and noise tolerance could be driven by variability in cognitive functions such as working memory, or musical experience. Studies using a reading span task (Daneman & Carpenter, 1980) have extensively investigated the association between working memory and NR processing (see review, Souza et al. (2015)). Although it is well-established that working memory capacity affects speech-in-noise performance (Akeroyd, 2008; Besser et al., 2013; Kim et al., 2020; Rönnberg et al., 2013; Rönnberg et al., 2008), the inconsistent interaction between working memory and NR processing was observed in explaining speech intelligibility or memory performance (see review, Souza et al. (2015)). Specifically, the literature found that listeners with larger working memory capacity tend to have better memory performance (Ng et al., 2013; Ng et al., 2015) but achieve worse speech intelligibility (Neher, 2014). In addition, the musical experience could lead to differences in supra-threshold coding fidelity that is not affected by cognitive variables (Varghese et al., 2015), resulting in differences in noise tolerance as well. Evidence indicated that the strength of the brainstem responses differs between musicians and non-musicians (Parbery-Clark et al., 2009; Wong et al., 2007). These results suggested that musical experience could be an important factor that influences the strength of encoding spectro-temporal features necessary for successful auditory object formation and, in turn, impacts listeners’ target-speech unmasking ability.

People with higher noise tolerance (i.e., group 2 in Figure 8) who had higher neural SNR got less NR benefits in behavioral accuracy compared to the other group of people with lower noise tolerance (i.e., group 1 in Figure 8). For high-tolerant listeners, NR processing seemed redundant and potentially harmful due to distortions for speech cues. Previous studies suggested that NR might be redundant, especially in laboratory settings in which participants were able to attend to auditory tasks fully, thus not improving speech-in-noise performance (Hafter & Schlauch, 1992). However, NR could be helpful in more natural settings requiring multiple attentional demands since it might do target-speech unmasking for listeners, relieving some cognitive load despite not improving speech-in-noise performance (e.g., Desjardins and Doherty (2014); Sarampalis et al. (2009)). Further, the current study results pointed out that NR might be redundant only for people with higher noise tolerance, who could unmask the target speech from background noise for themselves, suggesting the importance of examining individual noise tolerance in predicting potential NR benefits in each listener.

4.3. Methodological Validation, Limitations, and Implications

Our approach to combining EEG recording and NR processing in an experiment is worthy of a particular discussion. First, in terms of EEG recording, we considered using monosyllabic natural words, rather than sentences or speech syllables, to investigate cortical dynamics of speech-in-noise processing, particularly across the dorsal-lexicon regions (related to phonological discrimination) in conjunction with NR. For NR processing, however, we also needed to consider how long it took to be fully activated and how fast it could be engaged reacting to changes in estimated SNRs. Since HA features are applied in general gradually over a few seconds, previous HA studies (e.g., Desjardins and Doherty (2014)) used a sequence of sentences in the presence of background noise that started a few seconds before the onset of the first sentence and played continuously through the task. Here, we demonstrated that the NR algorithm in the current work was fully activated at around 200 ms after the noise onset and then reacted almost instantaneously to SNR changes in each time-point. Given that the target word played a half-second after the noise onset, these results ensured proper NR activation throughout the task. Thus, we were able to record EEG activity during the speech-in-noise tasks using monosyllabic natural words while securing the quality of NR processing.

The current study aimed to investigate behavioral and neural responses solely to NR processing and thus used a computer-based NR algorithm given potential challenges in isolating NR from other HA signal-processing schemes such as wide dynamic range compression (Wu & Stangl, 2013). In addition, it can be argued that findings from normal-hearing listeners could not generalize to HA users, although we first need to understand how the normal auditory and cognitive system works. A series of biology or device-driven factors (see review, Souza and Tremblay (2006)) ought to be considered for aided speech-in-noise processing in HA users (Billings et al., 2012; Billings et al., 2011; Billings et al., 2007; Tremblay et al., 2006). Future studies could consider using multiple-microphone NR, also known as “spatially-based NR,” estimating noise based on the spatial domain information (Metz, 2014; Popelka et al., 2018; Ricketts et al., 2019).

5. Conclusions

The present study indicates that NR could facilitate speech-in-noise processing, revealing direct, neural evidence of NR benefits occurring during the course of speech recognition. It is of note that this novel approach does not replace any current NR outcome measures but intends to add a mechanistic understanding of potential NR benefits. We also suggest a mechanism-based approach to individual NR outcomes by examining the cortical index of individual noise tolerance that could predict NR benefits in speech-in-noise performance. This biology-guided approach would likely provide more personalized NR configurations to individual needs.

Acknowledgement

The authors would like to thank Drs. Carolyn J. Brown, Shawn S. Goodman, and Camille C. Dunn for their constructive advice and valuable suggestions, and Ms. Ann Holmes for proofreading.

Conflicts of Interest and Source of Funding: This work was supported by the Council of Academic Programs in Communication Sciences and Disorders Ph.D. scholarship awarded to Kim, the Department of Defense Hearing Restoration Research Program Grant (W81XWH-19–1-0637) awarded to Choi, and NIH NIDCD P50 (DC000242 31). The authors declare no competing financial interests.

References

  1. Akeroyd MA (2008). Are individual differences in speech reception related to individual differences in cognitive ability? A survey of twenty experimental studies with normal and hearing-impaired adults. Int J Audiol, 47, 53–71. [DOI] [PubMed] [Google Scholar]
  2. Alcántara JI, Moore BCJ, Kühnel V, & Launer S. (2003). Evaluation of the noise reduction system in a commercial digital hearing aid: Evaluación del sistema de reducción de ruido en un auxiliar auditivo digital comercial. Int J Audiol, 42(1), 34–42. [DOI] [PubMed] [Google Scholar]
  3. Alexander JE, Bauer LO, Kuperman S, Morzorati S, O’Connor SJ, Rohrbaugh J, Porjesz B, Begleiter H, & Polich J. (1996). Hemispheric differences for P300 amplitude from an auditory oddball task. Int J Psychophysiol, 21(2), 189–196. [DOI] [PubMed] [Google Scholar]
  4. Ben-David BM, Chambers CG, Daneman M, Pichora-Fuller MK, Reingold EM, & Schneider BA (2011). Effects of aging and noise on real-time spoken word recognition: evidence from eye movements. J Speech Lang Hear Res, 54(1), 243–262. 10.1044/1092-4388(2010/09-0233) [DOI] [PubMed] [Google Scholar]
  5. Bentler R, Wu YH, Kettel J, & Hurtig R. (2008). Digital noise reduction: outcomes from laboratory and field studies. Int J Audiol, 47(8), 447–460. 10.1080/14992020802033091 [DOI] [PubMed] [Google Scholar]
  6. Bernarding C, Strauss DJ, Hannemann R, Seidler H, & Corona-Strauss FI (2013). Neural correlates of listening effort related factors: influence of age and hearing impairment. Brain Res Bull, 91, 21–30. 10.1016/j.brainresbull.2012.11.005 [DOI] [PubMed] [Google Scholar]
  7. Bernarding C, Strauss DJ, Hannemann R, Seidler H, & Corona-Strauss FI (2017). Neurodynamic evaluation of hearing aid features using EEG correlates of listening effort. Cogn Neurodyn, 11(3), 203–215. 10.1007/s11571-017-9425-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bernstein RS, & Gravel JS (1990, Jul). A method for determining hearing sensitivity in infants: the interweaving staircase procedure (ISP). J Am Acad Audiol, 1(3), 138–145. [PubMed] [Google Scholar]
  9. Besser J, Koelewijn T, Zekveld AA, Kramer SE, & Festen JM (2013). How linguistic closure and verbal working memory relate to speech recognition in noise-A review. Trends in Amplification, 17(2), 75–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Bharadwaj HM, Mai AR, Simpson JM, Choi I, Heinz MG, & Shinn-Cunningham BG (2019). Non-invasive assays of cochlear synaptopathy - candidates and considerations. Neuroscience, 407, 53–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Bharadwaj HM, Masud S, Mehraei G, Verhulst S, & Shinn-Cunningham BG (2015). Individual differences reveal correlates of hidden hearing deficits. J Neurosci, 35(5), 2161–2172. 10.1523/jneurosci.3915-14.2015 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Bharadwaj HM, Verhulst S, Shaheen L, Liberman MC, & Shinn-Cunningham BG (2014). Cochlear neuropathy and the coding of supra-threshold sound. Frontiers in systems neuroscience, 8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Billings CJ, McMillan GP, Penman TM, & Gille SM (2013). Predicting perception in noise using cortical auditory evoked potentials. JARO, 14(6), 891–903. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Billings CJ, Papesh MA, Penman TM, Baltzell LS, Gallun FJ, & Tremblay K. (2012). Clinical use of aided cortical auditory evoked potentials as a measure of physiological detection or physiological discrimination. International Journal of Otolaryngology, 2012, 1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Billings CJ, Penman TM, McMillan GP, & Ellis E. (2015). Electrophysiology and perception of speech in noise in older listeners: Effects of hearing impairment & age. Ear Hear, 36(6), 710–722. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Billings CJ, Tremblay KL, & Miller CW (2011). Aided cortical auditory evoked potentials in response to changes in hearing aid gain. Int J Audiol, 50(7), 459–467. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Billings CJ, Tremblay KL, Souza PE, & Binns MA (2007). Effects of hearing aid amplification and stimulus intensity on cortical auditory evoked potentials. Audiology and Neurotology, 12(4), 234–246. [DOI] [PubMed] [Google Scholar]
  18. Brainard DH (1997). The Psychophysics Toolbox. Spat Vis, 10(4), 433–436. https://www.ncbi.nlm.nih.gov/pubmed/9176952 [PubMed] [Google Scholar]
  19. Bregman AS (1990). Auditory scene analysis : the perceptual organization of sound. MIT Press. [Google Scholar]
  20. Bressler S, Goldberg H, & Shinn-Cunningham B. (2017). Sensory coding and cognitive processing of sound in Veterans with blast exposure. Hear Res, 349, 98–110. 10.1016/j.heares.2016.10.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Brons I, Dreschler WA, & Houben R. (2014). Detection threshold for sound distortion resulting from noise reduction in normal-hearing and hearing-impaired listeners. Journal of the Acoustical Society of America, 136(3), 1375–1384. [DOI] [PubMed] [Google Scholar]
  22. Caplan D, Gow D, & Makris N. (1995). Analysis of lesions by MRI in stroke patients with acoustic-phonetic processing deficits. Neurology, 45(2), 293–298. [DOI] [PubMed] [Google Scholar]
  23. Carcea I, Insanally MN, Froemke RC, Carcea I, Insanally MN, & Froemke RC (2017). Dynamics of auditory cortical activity during behavioural engagement and auditory perception. Nature Communications, 8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Choi I, Rajaram S, Varghese LA, & Shinn-Cunningham BG (2013). Quantifying attentional modulation of auditory-evoked cortical responses from single-trial electroencephalography. Front Hum Neurosci, 7, 115. 10.3389/fnhum.2013.00115 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Choi I, Wang L, Bharadwaj H, & Shinn-Cunningham B. (2014). Individual differences in attentional modulation of cortical responses correlate with selective attention performance. Hear Res, 314, 10–19. 10.1016/j.heares.2014.04.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Dai L, & Shinn-Cunningham BG (2016). Contributions of sensory coding and attentional control to individual differences in performance in spatial auditory selective attention tasks. Front Hum Neurosci, 10, 530. 10.3389/fnhum.2016.00530 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Dale AM, Liu AK, Fischl BR, Buckner RL, Belliveau JW, Lewine JD, & Halgren E. (2000). Dynamic statistical parametric mapping: combining fMRI and MEG for high-resolution imaging of cortical activity. Neuron, 26(1), 55–67. https://www.ncbi.nlm.nih.gov/pubmed/10798392 [DOI] [PubMed] [Google Scholar]
  28. Daneman M, & Carpenter PA (1980). Individual differences in working memory and reading. Journal of Verbal Learning & Verbal Behavior, 19(4), 450–466. 10.1016/S0022-5371(80)90312-6 [DOI] [Google Scholar]
  29. Darwin CJ (1997, Dec). Auditory grouping. Trends Cogn Sci, 1(9), 327–333. 10.1016/s1364-6613(97)01097-8 [DOI] [PubMed] [Google Scholar]
  30. Davis MH (2016). The Neurobiology of Lexical Access. In Hickok G. & Small SL (Eds.), Neurobiology of Language (pp. 541–555). Elsevier. [Google Scholar]
  31. Davis MH, & Johnsrude IS (2003). Hierarchical processing in spoken language comprehension. The Journal of Neuroscience, 23(8), 3423–3431. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. de Cheveigne A, & Nelken I. (2019). Filters: When, why, and how (not) to use them. Neuron, 102(2), 280–293. [DOI] [PubMed] [Google Scholar]
  33. Decruy L, Vanthornhout J, & Francart T. (2020). Hearing impairment is associated with enhanced neural tracking of the speech envelope. Hear Res, 393, 107961. [DOI] [PubMed] [Google Scholar]
  34. Desjardins JL, & Doherty KA (2014). The effect of hearing aid noise reduction on listening effort in hearing-impaired adults. Ear Hear, 35(6), 600–610. 10.1097/aud.0000000000000028 [DOI] [PubMed] [Google Scholar]
  35. Dillon H, & Lovegrove R. (1993). Single microphone noise reduction systems for hearing aids: A review and an evaluation. In Studebaker GA & Hochberg I. (Eds.), Acoustical factors affecting hearing aid performance (Vol. 20, pp. 353–372). Allyn and Bacon. [Google Scholar]
  36. Du Y, Buchsbaum BR, Grady CL, & Alain C. (2014). Noise differentially impacts phoneme representations in the auditory and speech motor systems. Proc Natl Acad Sci U S A, 111(19), 7126–7131. 10.1073/pnas.1318738111 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Du Y, Buchsbaum BR, Grady CL, & Alain C. (2016). Increased activity in frontal motor cortex compensates impaired speech perception in older adults. Nature Communications, 7, 12241. 10.1038/ncomms12241 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Ephraim Y, & Malah D. (1984). Speech enhancement using a minimum-mean square error short-time spectral amplitude estimator. IEEE Trans. Acoust., Speech, Signal Process, 32(6), 1109–1121. [Google Scholar]
  39. Ephraim Y, & Malah D. (1985). Speech enhancement using a minimum mean-square error log-spectral amplitude estimator. IEEE Trans. Acoust., Speech, Signal Process, 33(2), 443–445. [Google Scholar]
  40. Farris-Trimble A, & McMurray B. (2013). Test-retest reliability of eye tracking in the visual world paradigm for the study of real-time spoken word recognition. J Speech Lang Hear Res, 56(4), 1328–1345. 10.1044/1092-4388(2012/12-0145) [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Farris-Trimble A, McMurray B, Cigrand N, & Tomblin JB (2014). The process of spoken word recognition in the face of signal degradation. Journal of Experimental Psychology, 40(1), 308–327. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Fischl B, Sereno MI, Tootell RBH, & Dale AM (1999). High-resolution intersubject averaging and a coordinate system for the cortical surface. Hum Brain Mapp, 8(4), 272–284. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Friederici AD (2015). White-matter pathways for speech and language processing. Handb Clin Neurol, 129, 177–186. 10.1016/B978-0-444-62630-1.00010-X [DOI] [PubMed] [Google Scholar]
  44. Friston K, Harrison L, Daunizeau J, Kiebel S, Phillips C, Trujillo-Barreto N, Henson R, Flandin G, & Mattout J. (2008). Multiple sparse priors for the M/EEG inverse problem. Neuroimage, 39(3), 1104–1120. 10.1016/j.neuroimage.2007.09.048 [DOI] [PubMed] [Google Scholar]
  45. Geller J, McMurray B, Choi I, Holmes A, Schwalje A, Berger J, & Gander P. (2020). Validating the Iowa Test of Consonant Perception. PsyArXiv. 10.31234/osf.io/wxd93 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Gilmore CS, Clementz BA, & Berg P. (2009). Hemispheric differences in auditory oddball responses during monaural versus binaural stimulation. International Journal of Psychophysiology, 73(3), 326–333. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Giraud AL, & Poeppel D. (2012). Cortical oscillations and speech processing: emerging computational principles and operations. Nature neuroscience, 15(4), 511–517. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Gow DW Jr. (2012). The cortical organization of lexical knowledge: a dual lexicon model of spoken language processing. Brain Lang, 121(3), 273–288. 10.1016/j.bandl.2012.03.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Gow DW Jr., & Caplan D. (1996). An examination of impaired acoustic-phonetic processing in aphasia. Brain Lang, 52(2), 386–407. [DOI] [PubMed] [Google Scholar]
  50. Gramfort A, Luessi M, Larson E, Engemann DA, Strohmeier D, Brodbeck C, Goj R, Jas M, Brooks T, Parkkonen L, & Hamalainen M. (2013). MEG and EEG data analysis with MNE-Python. Front Neurosci, 7, 267. 10.3389/fnins.2013.00267 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Gramfort A, Luessi M, Larson E, Engemann DA, Strohmeier D, Brodbeck C, Parkkonen L, & Hamalainen MS (2014). MNE software for processing MEG and EEG data. Neuroimage, 86, 446–460. 10.1016/j.neuroimage.2013.10.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Gustafson S, McCreery R, Hoover B, Kopun JG, & Stelmachowicz P. (2014). Listening effort and perceived clarity for normal-hearing children with the use of digital noise reduction. Ear Hear, 35(2), 183–194. 10.1097/01.aud.0000440715.85844.b8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Hafter E, & Schlauch R. (1992). Cognitive factors and selection of auditory listening bands. In Dancer A, Henderson D, Salvi RJ, & Hammernik RP (Eds.), Noise-Induced Hearing Loss. (pp. 303–310). B.C. Decker. [Google Scholar]
  54. Hagerman B, & Olofsson Å. (2004). A method to measure the effect of noise reduction algorithms using simultaneous speech and noise. Acta Acustica united with Acustica, 90, 356–361. [Google Scholar]
  55. Hämäläinen MS, & Sarvas J. (1989). Realistic conductivity geometry model of the human head for interpretation of neuromagnetic data. IEEE Trans Biomed Eng, 36(2), 165–171. [DOI] [PubMed] [Google Scholar]
  56. Hickok G, & Poeppel D. (2007). The cortical organization of speech processing. Nat Rev Neurosci, 8(5), 393–402. 10.1038/nrn2113 [DOI] [PubMed] [Google Scholar]
  57. Hillyard SA, Hink RF, Schwent VL, & Picton TW (1973). Electrical signs of selective attention in the human brain. Science, 182(4108), 177–180. [DOI] [PubMed] [Google Scholar]
  58. Hillyard SA, Vogel EK, & Luck SJ (1998). Sensory gain control (amplification) as a mechanism of selective attention: electrophysiological and neuroimaging evidence. Philos Trans R Soc Lond B Biol Sci, 353(1373), 1257–1270. 10.1098/rstb.1998.0281 [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Hoetink AE, Korossy L, & Dreschler WA (2009). Classification of steady state gain reduction produced by amplitude modulation based noise reduction in digital hearing aids. Int J Audiol, 48(7), 444–455. 10.1080/14992020902725539 [DOI] [PubMed] [Google Scholar]
  60. Hornsby BW (2013). The effects of hearing aid use on listening effort and mental fatigue associated with sustained speech processing demands. Ear Hear, 34(5), 523–534. 10.1097/AUD.0b013e31828003d8 [DOI] [PubMed] [Google Scholar]
  61. Hsu JC (1996). Multiple comparisons : theory and methods. Chapman & Hall. [Google Scholar]
  62. Huettig F, & Altmann GT (2005). Word meaning and the control of eye fixation: semantic competitor effects and the visual world paradigm. Cognition, 96(1), B23–32. 10.1016/j.cognition.2004.10.003 [DOI] [PubMed] [Google Scholar]
  63. Humes LE (1999). Dimensions of hearing aid outcome. J Am Acad Audiol, 10(1), 26–39. [PubMed] [Google Scholar]
  64. Jamieson DG, Brennan RL, & Cornelisse LE (1995). Evaluation of a speech enhancement strategy with normal-hearing and hearing-impaired listeners. Ear Hear, 16(3), 274–286. 10.1097/00003446-199506000-00004 [DOI] [PubMed] [Google Scholar]
  65. Jung TP, Makeig S, Humphries C, Lee TW, McKeown MJ, Iragui V, & Sejnowski TJ (2000). Removing electroencephalographic artifacts by blind source separation. Psychophysiology, 37(2), 163–178. [PubMed] [Google Scholar]
  66. Kates J, & Arehart K. (2005). Multichannel Dynamic-Range Compression Using Digital Frequency Warping. Journal on Advances in Signal Processing, 2005. 10.1155/ASP.2005.3003 [DOI] [Google Scholar]
  67. Kates JM (2008). Digital hearing aids. Plural Pub. [Google Scholar]
  68. Kim S, Choi I, Schwalje AT, Kim K, & Lee JH (2020). Auditory working memory explains variance in speech recognition in older listeners under adverse listening conditions. Clinical interventions in aging, 15, 395–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Kim S, Schwalje AT, Liu AS, Gander PE, McMurray B, Griffiths TD, & Choi I. (2021). Pre- and post-target cortical processes predict speech-in-noise performance. Neuroimage, 228, 117699. 10.1016/j.neuroimage.2020.117699 [DOI] [PMC free article] [PubMed] [Google Scholar]
  70. Levitt H. (2001). Noise reduction in hearing aids : a review. Journal of Rehabilitation Research and Development, 38(1), 111–121. [PubMed] [Google Scholar]
  71. Lewis JD, Goodman SS, & Bentler RA (2010). Measurement of hearing aid internal noise. J Acoust Soc Am, 127(4), 2521–2528. 10.1121/1.3327808 [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Liberman MC, Epstein MJ, Cleveland SS, Wang H, & Maison SF (2016). Toward a differential diagnosis of hidden hearing loss in humans. PLoS One, 11(9), 1–16. 10.1371/journal.pone.0162726 [DOI] [PMC free article] [PubMed] [Google Scholar]
  73. Luck SJ (2014). An introduction to the event-related potential technique. The MIT Press. [Google Scholar]
  74. Ma J, & Loizou PC (2011). SNR Loss: A new objective measure for predicting speech intelligibility of noise-suppressed speech. Speech communication, 53(3), 340–354. 10.1016/j.specom.2010.10.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Mai A, Flesher B, Dougherty K, Hagedorn A, Simpson JM, Heinz MG, & Bharadwaj HM (2019). Physiological assays of suprathreshold hearing are consistent with widespread deafferentation of the human auditory periphery. J Acoust Soc Am, 145(3), 1663. [Google Scholar]
  76. Makeig S, Bell AJ, Jung T-P, & Sejnowski TJ (1996). Independent component analysis of electroencephalographic data. Advances in neural information processing systems, 145–151. [Google Scholar]
  77. Maxwell SE, & Delaney HD (2004). Designing experiments and analyzing data a model comparison perspective. Lawrence Erlbaum Associates. [Google Scholar]
  78. McMurray B, Farris-Trimble A, & Rigler H. (2017). Waiting for lexical access: Cochlear implants or severely degraded input lead listeners to process speech less incrementally. Cognition, 169, 147–164. 10.1016/j.cognition.2017.08.013 [DOI] [PMC free article] [PubMed] [Google Scholar]
  79. Mesgarani N, & Chang EF (2012). Selective cortical representation of attended speaker in multi-talker speech perception. Nature, 485(7397), 233–236. 10.1038/nature11020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  80. Metz MJ (2014). Sandlin’s textbook of hearing aid amplification : technical and clinical considerations. Plural Publishing. [Google Scholar]
  81. Mueller HG, Weber J, & Hornsby BW (2006). The effects of digital noise reduction on the acceptance of background noise. Trends Amplif, 10(2), 83–93. 10.1177/1084713806289553 [DOI] [PMC free article] [PubMed] [Google Scholar]
  82. Naylor G, & Johannesson RB (2009). Long-term signal-to-noise ratio at the input and output of amplitude-compression systems. J Am Acad Audiol, 20(3), 161–171. [DOI] [PubMed] [Google Scholar]
  83. Neher T. (2014). Relating hearing loss and executive functions to hearing aid users’ preference for, and speech recognition with, different combinations of binaural noise reduction and microphone directionality. Front Neurosci, 8, 391. 10.3389/fnins.2014.00391 [DOI] [PMC free article] [PubMed] [Google Scholar]
  84. Neher T, Grimm G, Hohmann V, & Kollmeier B. (2014). Do hearing loss and cognitive function modulate benefit from different binaural noise-reduction settings? Ear Hear, 35(3), e52–62. 10.1097/aud.0000000000000003 [DOI] [PubMed] [Google Scholar]
  85. Neher T, & Wagener KC (2016). Investigating differences in preferred noise reduction strength among hearing aid users. Trends in hearing, 20, 1–14. 10.1177/2331216516655794 [DOI] [PMC free article] [PubMed] [Google Scholar]
  86. Neher T, Wagener KC, & Fischer RL (2016). Directional processing and noise reduction in hearing aids: Individual and situational influences on preferred setting. J Am Acad Audiol, 27(8), 628–646. 10.3766/jaaa.15062 [DOI] [PubMed] [Google Scholar]
  87. Ng EH, Rudner M, Lunner T, Pedersen MS, & Ronnberg J. (2013, Jul). Effects of noise and working memory capacity on memory processing of speech for hearing-aid users. Int J Audiol, 52(7), 433–441. 10.3109/14992027.2013.776181 [DOI] [PubMed] [Google Scholar]
  88. Ng EH, Rudner M, Lunner T, & Rönnberg J. (2015). Noise reduction improves memory for target language speech in competing native but not foreign language speech. Ear Hear, 36(1), 82–91. [DOI] [PubMed] [Google Scholar]
  89. Nunez PL, & Westdorp AF (1994). The surface Laplacian, high resolution EEG and controversies. Brain Topogr, 6(3), 221–226. 10.1007/bf01187712 [DOI] [PubMed] [Google Scholar]
  90. O’Sullivan J, Mesgarani N, Smith E, Schevon C, McKhann GM, Sheth SA, Herrero J, Mehta AD, Smith E, & Sheth SA (2019). Hierarchical encoding of attended auditory objects in multi-talker speech perception. Neuron, 104(6), 1195–1209.e1193. [DOI] [PMC free article] [PubMed] [Google Scholar]
  91. Obleser J, & Erb J. (2020). Neural filters for challenging listening situations. In Poeppel D, Mangun GR, Gazzaniga MS, & Preuss TM (Eds.), The Cognitive Neurosciences (6 ed., pp. 167–176). MIT Press. [Google Scholar]
  92. Ohlenforst B, Wendt D, Kramer SE, Naylor G, Zekveld AA, & Lunner T. (2018). Impact of SNR, masker type and noise reduction processing on sentence recognition performance and listening effort as indicated by the pupil dilation response. Hear Res, 365, 90–99. [DOI] [PubMed] [Google Scholar]
  93. Ohlenforst B, Zekveld AA, Jansma EP, Wang Y, Naylor G, Lorens A, Lunner T, & Kramer SE (2017). Effects of hearing impairment and hearing aid amplification on listening effort: A systematic review. Ear Hear, 38(3), 267–281. 10.1097/AUD.0000000000000396 [DOI] [PMC free article] [PubMed] [Google Scholar]
  94. Parbery-Clark A, Marmel F, Bair J, & Kraus N. (2011). What subcortical-cortical relationships tell us about processing speech in noise. European Journal of Neuroscience, 33(3), 549–557. [DOI] [PubMed] [Google Scholar]
  95. Parbery-Clark A, Skoe E, & Kraus N. (2009). Musical experience limits the degradative effects of background noise on the neural processing of sound. J Neurosci, 29(45), 14100–14107. 10.1523/JNEUROSCI.3256-09.2009 [DOI] [PMC free article] [PubMed] [Google Scholar]
  96. Pelli DG (1997). The VideoToolbox software for visual psychophysics: transforming numbers into movies. Spat Vis, 10(4), 437–442. https://www.ncbi.nlm.nih.gov/pubmed/9176953 [PubMed] [Google Scholar]
  97. Pisa J, Burk M, & Galster EJTHJ (2010). Evidence-based design of a noise-management algorithm. 63(4), 42–44. [Google Scholar]
  98. Plomp R. (1994). Noise, amplification, and compression: considerations of three main issues in hearing aid design. Ear Hear, 15(1), 2–12. [PubMed] [Google Scholar]
  99. Popelka GR, Moore BCJ, Fay RR, & Popper AN (2018). Hearing Aids. Springer International Publishing, A. G. [Google Scholar]
  100. Rauschecker JP, & Scott SK (2009). Maps and streams in the auditory cortex: nonhuman primates illuminate human speech processing. Nature neuroscience, 12(6), 718–724. [DOI] [PMC free article] [PubMed] [Google Scholar]
  101. Ricketts T, Bentler RA, & Mueller HG (2019). Essentials of modern hearing aids: selection, fitting, and verification. Plural Publishing. [Google Scholar]
  102. Ricketts T, & Hornsby B. (2005). Sound quality measures for speech in noise through a commercial hearing aid implementing digital noise reduction. J Am Acad Audiol, 16(5), 270–277. 10.3766/jaaa.16.5.2 [DOI] [PubMed] [Google Scholar]
  103. Rönnberg J, Lunner T, Zekveld A, Sörqvist P, Danielsson H, Lyxell B, Dahlström O, Signoret C, Stenfelt S, Pichora-Fuller MK, & Rudner M. (2013). The ease of language understanding (ELU) model: theoretical, empirical, and clinical advances. Frontiers in systems neuroscience, 7:31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  104. Rönnberg J, Rudner M, Foo C, & Lunner T. (2008). Cognition counts: A working memory system for ease of language understanding (ELU). Int J Audiol, 47(sup2), S99–S105. [DOI] [PubMed] [Google Scholar]
  105. Ruggles D, Bharadwaj H, & Shinn-Cunningham BG (2011). Normal hearing is not enough to guarantee robust encoding of suprathreshold features important in everyday communication. Proc Natl Acad Sci U S A, 108(37), 15516–15521. 10.1073/pnas.1108912108 [DOI] [PMC free article] [PubMed] [Google Scholar]
  106. Sarampalis A, Kalluri S, Edwards B, & Hafter E. (2009). Objective measures of listening effort: effects of background noise and noise reduction. J Speech Lang Hear Res, 52(5), 1230–1240. 10.1044/1092-4388(2009/08-0111) [DOI] [PubMed] [Google Scholar]
  107. Satterthwaite FE (1946). An approximate distribution of estimates of variance components. Biometric Bulletin, 2, 110–114. [PubMed] [Google Scholar]
  108. Souza P, Arehart K, & Neher T. (2015). Working memory and hearing aid processing: Literature findings, future directions, and clinical applications. Frontiers in psychology, 6, 1894–1894. 10.3389/fpsyg.2015.01894 [DOI] [PMC free article] [PubMed] [Google Scholar]
  109. Souza PE, & Tremblay KL (2006). New perspectives on assessing amplification effects. Trends in Amplification, 10(3), 119–143. 10.1177/1084713806292648 [DOI] [PMC free article] [PubMed] [Google Scholar]
  110. Srinivasan R, Nunez PL, Tucker DM, Silberstein RB, & Cadusch PJ (1996). Spatial sampling and filtering of EEG with spline laplacians to estimate cortical potentials. Brain Topogr, 8(4), 355–366. 10.1007/bf01186911 [DOI] [PubMed] [Google Scholar]
  111. Stelmachowicz P, Lewis D, Hoover B, Nishi K, McCreery R, & Woods W. (2010). Effects of Digital Noise Reduction on Speech Perception for Children with Hearing Loss. Ear Hear, 31, 345–355. 10.1097/AUD.0b013e3181cda9ce [DOI] [PMC free article] [PubMed] [Google Scholar]
  112. Strauss DJ, Corona-Strauss FI, Trenado C, Bernarding C, Reith W, Latzel M, & Froehlich M. (2010). Electrophysiological correlates of listening effort: neurodynamical modeling and measurement. Cogn Neurodyn, 4(2), 119–131. 10.1007/s11571-010-9111-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  113. Studebaker GA (1985). A “rationalized” arcsine transform. Journal of speech and hearing research, 28(3), 455–462. [DOI] [PubMed] [Google Scholar]
  114. Takahashi G, Martinez CD, Beamer S, Bridges J, Noffsinger D, Sugiura K, Bratt GW, & Williams DW (2007). Subjective measures of hearing aid benefit and satisfaction in the NIDCD/VA follow-up study. J Am Acad Audiol, 18(4), 323–349. [DOI] [PubMed] [Google Scholar]
  115. Teki S, Barascud N, Picard S, Payne C, Griffiths TD, & Chait M. (2016). Neural correlates of auditory figure-ground segregation based on temporal coherence. Cereb Cortex, 26(9), 3669–3680. 10.1093/cercor/bhw173 [DOI] [PMC free article] [PubMed] [Google Scholar]
  116. Teki S, Chait M, Kumar S, Shamma S, & Griffiths TD (2013). Segregation of complex acoustic scenes based on temporal coherence. Elife, 2, e00699. 10.7554/eLife.00699 [DOI] [PMC free article] [PubMed] [Google Scholar]
  117. Tong Y, Chen Q, Nichols TE, Rasetti R, Callicott JH, Berman KF, Weinberger DR, & Mattay VS (2016). Seeking optimal region-of-interest (ROI) single-value summary measures for fMRI studies in imaging genetics. PLoS One, 11(3), 1–20. 10.1371/journal.pone.0151391 [DOI] [PMC free article] [PubMed] [Google Scholar]
  118. Tremblay KL, Billings CJ, Friesen LM, & Souza PE (2006). Neural Representation of Amplified Speech Sounds. Ear Hear, 27(2), 93–103. [DOI] [PubMed] [Google Scholar]
  119. Vanthornhout J, Decruy L, Wouters J, Simon JZ, & Francart T. (2018). Speech Intelligibility Predicted from Neural Entrainment of the Speech Envelope. JARO, 19(2), 181–191. [DOI] [PMC free article] [PubMed] [Google Scholar]
  120. Varghese L, Bharadwaj HM, & Shinn-Cunningham BG (2015). Evidence against attentional state modulating scalp-recorded auditory brainstem steady-state responses. Brain Research, 1626, 146–164. [DOI] [PMC free article] [PubMed] [Google Scholar]
  121. Viswanathan V, Bharadwaj HM, & Shinn-Cunningham BG (2019). Electroencephalographic signatures of the neural representation of speech during selective attention. eNeuro, 6(5). [DOI] [PMC free article] [PubMed] [Google Scholar]
  122. Ward KM, Shen J, Souza PE, & Grieco-Calub TM (2017). Age-related differences in listening effort during degraded speech recognition. Ear Hear, 38(1), 74–84. 10.1097/aud.0000000000000355 [DOI] [PMC free article] [PubMed] [Google Scholar]
  123. Wendt D, Hietkamp RK, & Lunner T. (2017). Impact of noise and noise reduction on processing effort: A pupillometry study. Ear Hear, 38(6), 690–700. 10.1097/aud.0000000000000454 [DOI] [PubMed] [Google Scholar]
  124. Wild CJ, Yusuf A, Wilson DE, Peelle JE, Davis MH, & Johnsrude IS (2012). Effortful listening: the processing of degraded speech depends critically on attention. J Neurosci, 32(40), 14010–14021. 10.1523/JNEUROSCI.1528-12.2012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  125. Winn MB, Edwards JR, & Litovsky RY (2015). The impact of auditory spectral resolution on listening effort revealed by pupil dilation. Ear Hear, 36(4), e153–e165. 10.1097/AUD.0000000000000145 [DOI] [PMC free article] [PubMed] [Google Scholar]
  126. Woldorff MG, Gallen CC, Hampson SA, Hillyard SA, Pantev C, Sobel D. Bloom FE (1993). Modulation of early sensory processing in human auditory cortex during auditory selective attention. Proc Natl Acad Sci U S A, 90(18), 8722–8726. [DOI] [PMC free article] [PubMed] [Google Scholar]
  127. Wong PC, Jin JX, Gunasekera GM, Abel R, Lee ER, & Dhar S. (2009). Aging and cortical mechanisms of speech perception in noise. Neuropsychologia, 47(3), 693–703. 10.1016/j.neuropsychologia.2008.11.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
  128. Wong PC, Skoe E, Russo NM, Dees T, & Kraus N. (2007). Musical experience shapes human brainstem encoding of linguistic pitch patterns. Nature neuroscience, 10(4), 420–422. [DOI] [PMC free article] [PubMed] [Google Scholar]
  129. Wu YH, & Bentler RA (2007). Using a signal cancellation technique to assess adaptive directivity of hearing aids. J Acoust Soc Am, 122(1), 496–511. 10.1121/1.2735804 [DOI] [PubMed] [Google Scholar]
  130. Wu YH, & Bentler RA (2009). Using a signal cancellation technique involving impulse response to assess directivity of hearing aids. J Acoust Soc Am, 126(6), 3214–3226. 10.1121/1.3238161 [DOI] [PubMed] [Google Scholar]
  131. Wu YH, & Bentler RA (2011). A method to measure hearing aid directivity index and polar pattern in small and reverberant enclosures. Int J Audiol, 50(6), 405–416. 10.3109/14992027.2010.551219 [DOI] [PubMed] [Google Scholar]
  132. Wu YH, & Stangl E. (2013). The effect of hearing aid signal-processing schemes on acceptable noise levels: perception and prediction. Ear Hear, 34(3), 333–341. 10.1097/AUD.0b013e31827417d4 [DOI] [PubMed] [Google Scholar]

RESOURCES