A critical review of hearing-aid single-microphone noise-reduction studies in adults and children

Abstract

This evidence supports the use of SMNR for adults and school-aged children when the aim is to improve listening comfort or reduce listening effort. Future research should test SMNR with infants and children who are younger than 5 years of age. Further development, testing, and clinical trials should be carried out on algorithms not yet available in wearable hearing aids. Testing higher cognitive level for speech processing and learning of novel sounds or words could show benefits of advanced signal processing features. These approaches should be expanded to other populations such as children and younger adults. Implications for rehabilitation The review provides a quick reference for students and clinicians regarding the efficacy and effectiveness of SMNR in wearable hearing aids. This information is useful during counseling session to build a realistic expectation among hearing aid users. Most studies in the adult population suggest that SMNR may provide some benefits to adult listeners in terms of listening comfort, acceptance of background noise, and release of cognitive load in a complex listening condition. However, it does not improve speech intelligibility. Studies that examined SMNR in the paediatric population suggest that SMNR may benefit older school-aged children, aged between 10 and 12 years old. The evidence supports the use of SMNR for adults and school-aged children when the aim is to improve listening comfort or reduce listening effort.

Output signal-to-noise ratio and speech perception in noise: effects of algorithm

“…The modifications to SNR in the current study depended on the HA device and type of processing, with WDRC showing the greatest effect on SNR. The differences between previous results (e.g., Hagerman & Olofsson, 2004; Souza et al, 2006; Naylor & Johannesson, 2009; Chong & Jenstad, 2010; Wu & Stangl, 2013; Alexander & Masterson, 2015; and Brons et al, 2015) and the current results are likely due to differences in test parameters, such as input SNR and type of background noise. The input SNR varied based on individual needs (i.e., SRTn), which could have concealed more systematic effects had multiple fixed SNRs been tested across the psychometric function.…”

“…Several studies have used the phase-inversion approach proposed by Hagerman and Olofsson to quantify acoustic changes to speech and noise post-HA processing, including estimation of the SNR of the HA’s output (Hagerman & Olofsson, 2004; Olsen et al, 2005; Souza et al, 2006; Naylor & Johannesson, 2009; Chong & Jenstad, 2010; Gustafson et al, 2014; Alexander & Masterson, 2015). Using the phase-inversion technique, researchers have demonstrated that WDRC can result in either increases or decreases of the long-term output SNR, depending on both the input SNR and the HA compression settings (Olsen et al, 2005; Souza et al, 2006; Naylor & Johannesson, 2009).…”

Signal processing in auditory cortex underlies degraded speech sound discrimination in noise

“…The challenges of hearing in noise are particularly problematic for individuals with hearing loss or deafness and those using hearing aids or cochlea implants, who receive limited benefits from noise-reduction techniques (Brons, Houben et al 2012, Chong andJenstad 2018). Approaches to reduce noise often attempt to characterize and subtract background sounds from incoming signals, essentially mimicking the effects of adaptation observed in neural responses to ongoing sounds.…”

Using the phase inversion method and loudness comparisons for the evaluation of noise reduction algorithms in hearing aids

“…Various studies with test subjects or technical measurements for the evaluation of a NRSE can be found in literature [7] which show that a NRSE mostly does not increase speech intelligibility (e.g., [8][9][10][11][12]) but can reduce listening effort. To determine the impact on listening effort, various approaches have been reported, e.g., measuring pupil dilation [13,14], electroencephalography (EEG) [15,16], electrodermal activity [17], response time [18,19], using a dual task paradigm [20][21][22][23][24], or asking for subjective ratings, e.g., using the acceptable noise level (ANL) [25,26], questionnaires [27], or adaptive scaling procedures [28].…”