Age-related sensorineural hearing loss is often associated with a bilateral, downward-sloping audiometric configuration that results in limited access to high-frequency sounds (Ear Hear. 2003 Jun;24(3):198). This, in turn, can cause restricted audible bandwidth that may negatively impact speech perception (Ear Hear. 2013 Mar;34(2):e24). Hearing aids are the most common treatment for hearing loss, but these devices may not be able to restore audibility at high frequencies due to the severity of a person's hearing loss. To make high-frequency cues available to people with restricted high-frequency hearing, non-linear frequency compression (NLFC) was developed as a strategy to shift high-frequency information to lower, more audible frequency regions. This shift is often adjusted within a hearing aid manufacturer's software, in which the cut-off frequency and compression ratio are adjusted based on a patient's audiogram. Despite the assumption that high-frequency cues are restored by NLFC, the benefits seen from this strategy have been inconsistent. Some studies did not find any benefit in NLFC (Ear Hear. 2014 Jul-Aug;35(4):440). Numerous studies have investigated the performance of individuals with NLFC in recognizing consonants, words, and sentences in quiet, but only a few have investigated the benefits of NFLC in the speech perception in noise of older adults using commercially available hearing aids.
SPEECH PERCEPTION IN NOISE WITH NLFC
In 2015, Ellis and Munro published two studies that measured speech perception in quiet and in noise of an older adult population using commercially available hearing aids (Int J Audiol. 2015 Jan;54(1):37; Int J Audiol. 2015 Jul;54(7):467). NLFC resulted in significantly improved recognition of nonsense syllables in quiet and in babble, as well as significantly improved sentence recognition in babble. Participants with poorer high-frequency pure tone averages experienced more benefits from NLFC, presumably due to the restoration of high-frequency cues.
However, two other studies that tested a similar population did not find the same perceptual benefits. Hopkins and colleagues found significant improvement in consonant recognition but none in sentence recognition in noise among participants with poor high-frequency thresholds who used NLFC hearing aids (Int J Audiol. 2014 Apr;53(4):219). Similarly, Picou and colleagues did not find any NLFC benefit in sentence recognition in noise, although their test population only had mild to moderate hearing loss (Int J Audiol. 2015 Mar;54(3):162). They noted significant inter-individual variability in the benefits of this hearing aid strategy.
Despite these variable study findings, several hearing aid manufacturers automatically enable their own version of NLFC during a “first-fit” procedure that makes use of a patient's audiogram to determine the strategy's settings. Although there is evidence of improved consonant recognition among individuals with greater degrees of high-frequency hearing loss, there is not enough evidence to suggest that all or even most patients would experience speech recognition benefits in quiet or in noise, making a default-on approach unsuitable. It is critical to further examine the effects of NLFC on speech recognition in noise and the relationship between outcome and listener variables to identify the factors associated with any benefit. Speech recognition measures alone may not capture other potential benefits from a hearing aid strategy such as reduced listening effort. Several studies have observed reduced listening effort due to digital noise reduction (e.g., Ear Hear. 2014 Nov-Dec;35(6):600; J Speech Lang Hear Res. 2009 Oct;52(5):1230), but previous studies on NLFC have not measured this domain.
In a recently published study in Ear and Hearing, we reported the results on speech perception in noise and listening effort with NLFC hearing aids (Ear Hear. 2018 Mar/Apr;39(2):215). We recruited 17 adults (age: 57-85) with mild to moderately severe sensorineural hearing loss. While wearing hearing aids fit to NAL-NL2 gain targets with or without NLFC enabled, our participants were tasked to repeat the last words of sentences that were either rich or neutral in context. The sentences were presented in babble from a front loudspeaker, and the signal-to-noise ratio (SNR) was set to 50 percent of each participant's SNR threshold to avoid ceiling and floor effects. To measure listening effort, we gave the participants a secondary task to repeat as many of the sentence-ending words as they could recall after a block of six sentences. We were also interested in determining whether individual differences in hearing and cognition were associated with the amount of benefit from NLFC. Our hearing factors were hearing loss severity, hearing loss slope, and audible bandwidth with NLFC. For cognition we measured the working memory capacity, which is often measured in studies on cognition and speech perception.
EFFECTS OF NON-LINEAR FREQUENCY COMPRESSION
We observed improved recognition in noise and reduced listening effort with NLFC enabled, which suggested that NLFC may help speech recognition on average in noisy listening environments. It is likely that NLFC restored high-frequency speech acoustic cues, improved speech perception, and reduced the effort necessary for that level of perception. While the average improvement was only about five percent in both recognition and listening effort, outcomes varied among our participants. Some participants had largely improved performance, while the performance of others remained relatively unchanged. Most of our participants’ speech recognition was either improved or unchanged, and only a few participants’ performance declined, suggesting that a default-on approach to hearing aid settings may not be as detrimental as anticipated.
To better understand this variability, we measured several auditory and cognitive factors as potential predictors of NLFC benefit. We found that individuals with more steeply sloping hearing loss received greater benefits from NLFC. This confirms the findings of previous studies that positively correlated NLFC benefit with high-frequency hearing loss, again suggesting that this hearing aid strategy successfully restores high-frequency cues. Also of note is that prior studies that did not find recognition benefits from NLFC used the NAL-NL1 gain prescription, whereas our study and those by Ellis and Munro used the NAL-NL2 gain prescription that typically provides more high-frequency gain than NAL-NL1. It is possible that the prior version of gain prescription did not provide enough gain at high frequencies, resulting in fewer high-frequency cues that are compressed.
We also observed improved word recall performance in this complex listening task with NLFC enabled. Recall benefit due to NLFC was negatively associated with age. Older participants received less recall benefit than younger participants in the older adult cohort. Working memory capacity was surprisingly not a predictor of either recognition or recall benefits in noise. Since age was a significant predictor of recall benefit and working memory capacity was not, it is possible that our measure of working memory capacity did not fully capture the age-related cognitive changes that have been associated with speech recognition in noise and hearing aid strategy benefit.
Recognition and recall benefits were more pronounced in context-neutral sentences, which suggested that participants successfully used available context cues to determine the sentence-ending word. However, when the sentence context was neutral, the participants were forced to rely on the amplified and processed acoustic signal, rendering the additional acoustic cues provided by NLFC to be beneficial. While this effect could limit the potential benefit of NLFC to listening scenarios that lack context, it is not uncommon for noisy listening environments to have limited context, especially for individuals with hearing loss. Additionally, potential benefit cannot be ruled out in context-rich listening situations since many of our participants were near ceiling performance for context-rich sentences.
To conclude, our study found that hearing aids with NLFC improved speech recognition in noise and reduced listening effort. Individuals with steeply sloping hearing loss are more likely to benefit from NLFC, although this benefit may be limited by aging effects.