At Hearts for Hearing, our pediatric audiologists are experts in the art and science of ensuring children with hearing loss have consistent audibility of speech and other important environmental sounds. Ensuring audibility of fricative speech sounds like /s/, /θ/ (the voiceless “th” as in the word “thin”), and /f/ can be a challenge when fitting hearing aids in children with significant high-frequency hearing loss. In some cases, audiologists may need to enable frequency-lowering processing in hearing aids to optimize audibility of high-frequency sounds. If baby Garth were fitted with hearing aids with frequency-lowering technology, he might belt out that he has fricatives in low places. In this month's installment of Tot 10, we explore frequency-lowering processing for children with hearing loss.
10. Why would we consider frequency-lowering hearing technology for infants and children with hearing loss?
Frequency-lowering processing for hearing aids was introduced around 2005. Today, this technology is offered by all major hearing aid manufacturers. Hearing scientists and audiologists have long recognized the contribution of the high-frequency portion of the speech spectrum (i.e., 2,000 Hz and above) to speech recognition. However, in the early 2000s, Stelmachowicz, et al., shared research results that upped the ante on the importance of the audibility of high-frequency speech sounds for infants and young children (J Acoust Soc Am. 2001 Oct;110(4):2183). Specifically, they showed that adults required a bandwidth extending to 6,000 Hz to reach optimal recognition of the phoneme /s/. In contrast, children reached optimal recognition of the /s/ phoneme when the bandwidth was extended to 9,000 Hz. Additionally, they reported that the /s/ phoneme possessed a center frequency of 6,000 Hz when spoken by a man, whereas the center frequency was closer to 9,000 Hz when produced by a woman or child. Moreover, the Boystown group has shown that many children with significant high-frequency hearing loss have deficits in their production of /s/, which is not surprising considering that children's speech articulation skills are dependent upon their ability to clearly hear the variety of speech sounds in their native tongue (Arch Otolaryngol Head Neck Surg. 2004 May;130(5):556).
9. When would we not want to consider frequency-lowering processing for infants and children with hearing loss?
By definition, distortion occurs when the output signal of a system is altered in shape or form relative to the input signal to the system. Frequency-lowering processing distorts sound by altering the input signal so that the output is different in form. If a substantial amount of frequency-lowering is applied, then the resultant distortion can lead to poor sound quality and possibly even a degradation in speech recognition. For example, the /s/ phoneme may become indistinguishable from the /sh/ phoneme or other fricatives. As a result, the audiologist must carefully select children who need frequency-lowering processing and the ideal parameters so that the technology improves audibility while curtailing the negative effects that may come with distortion. On the bright side, research has shown that when frequency-lowering technology is applied appropriately, its positive effects are realized and the negative effects are minimized.
8. Many children have varying degrees and configurations of hearing loss. Which ones will benefit from frequency-lowering technology?
Those who do not have satisfactory audibility of high-frequency speech sounds will possibly benefit from this technology. Conventional wisdom may suggest that frequency-lowering processing should only be enabled for children who have severe to profound high-frequency hearing loss. However, McCreery, et al., and Wolfe, et al., have shown that many children with mild to moderate hearing loss can benefit from frequency-lowering technology (Ear Hear. 2014 Jul-Aug;35(4):440; J Am Acad Audiol. 2010 Nov-Dec;21(10):618; Int J Audiol. 2011 Jun;50(6):39; Int J Audiol. 2015 Mar;54(3):170). Of note, research has also shown that some children do not exhibit improvements in speech recognition with the use of frequency-lowering processing, even when its use improves the audibility of high-frequency speech sounds (Int J Audiol. 2009 Sep; 48(1): 632). This lack of benefit may be attributed to a listener's poor spectral processing ability, limited working memory, and/or deficient cognitive skills. More research is needed to clarify the endogenous factors that may affect the benefit children get from this technology.
7. How should I determine whether I must use frequency-lowering processing?
There are several ways to determine whether frequency-lowering processing should be enabled, and if so, how the parameters should be set. The audiologist should begin by completing a real ear measurement to see if adequate audibility can be provided for the entire speech spectrum without the use of frequency-lowering technology. Figure 1 shows a sample output of a hearing aid fitted with frequency-lowering processing disabled (turned off). The hearing aid output is fitted to DSL 5.0 prescriptive targets for children, but audibility is limited above 3,000 Hz.
Experts from the University of Western Ontario (UWO) have developed calibrated speech sounds that may be presented from probe microphone systems to see if speech sounds are audible with and without frequency-lowering processing (AudiologyOnline. 2016;Article 16932). Figure 2 provides an example of hearing aid output in response to the phoneme /s/ presented at 65 dB SPL. The UWO speech stimuli contain the same acoustical characteristics as real speech sounds; consequently, the audiologist may observe how the hearing aid will respond to speech sounds encountered during real world use.
It is important to verify the output of the hearing aid with frequency-lowering processing disabled. As previously noted, frequency-lowering technology introduces at least some distortion to the audio signal, so if verification indicates sufficient audibility is provided without frequency-lowering processing, the audiologist should not activate the feature. Additionally, verification without frequency-lowering technology allows the audiologist to determine the limits of audibility, which will facilitate the identification of the frequency range that needs to be lowered.
6. What do I do if high-frequency speech sounds are not audible without frequency-lowering?
Again, the UWO team provides excellent counsel on the use of verification to determine the most appropriate starting parameters for frequency-lowering technology. First, the audiologist should identify the maximum audible output frequency (MAOF) range, defined by the range that exists between the point at which the long-term average speech spectrum (LTASS) and the peak of the speech signal cross the child's threshold line (i.e., the SPL-o-gram; see Fig. 3). Next, the audiologist should activate frequency-lowering and use the weakest setting that results in the upper shoulder of the aided /s/ (presented at 65 dB SPL) signal falling at the upper limit of the MAOF range (see Fig. 4).
5. That seems pretty simple. Is that all I need to do?
Audiologists can take additional steps to increase the likelihood that frequency-lowering processing will not cause deleterious effects. First, the audiologist can present the /∫/ speech sound (i.e., the fricative, “sh” as in the word, “sheet”) at 65 dB SPL and confirm there is no overlap between the /s/ and /∫/ speech sounds. The UWO group recommends at least a 1/3 octave band of separation between the lower shoulders of the /s/ and /∫/ speech sounds. Of note, the tick marks on the x-axis of Figure 5 represent 1/3 octave bands; the separation between the lower shoulder of /s/ and /∫/ is much greater than 1/3 octave.
4. What other steps can I take to optimize performance with frequency-lowering processing?
First, the audiologist should always conduct a simple listening check to evaluate the effect of frequency-lowering processing on sound quality. Many real ear probe microphone systems allow the audiologist to connect earphones to the system to listen to the hearing aid output. If the audiologist has normal auditory function and cannot distinguish between the /s/ and /∫/ phonemes, it is very unlikely that the child will be able to discriminate between the two phonemes. In such a case, the audiologist may consider providing weaker frequency-lowering processing settings. Additionally, for children who have severe to profound hearing loss in the mid- to high-frequency range (1,000 to 3,000 Hz), the audiologist should confirm that the frequency-lowering is not so strong that it alters the second formant frequency of vowels resulting in speech recognition difficulties. The audiologist may want to consider using the live voice feature of the probe microphone system to examine the spectral peaks of vowels (e.g., consider beginning with the vowel /i/, as in the “ee” in she) to ensure the second formant has not been altered.
Furthermore, the audiologist should administer validation measures to evaluate the effect of frequency-lowering processing. The UWO group has developed several measures that may be used to assess the potential benefit of frequency-lowering technology. For instance, the norm-referenced Ling 6 Sounds may be administered using an audiometer (via a CD or computer) to determine a child's aided thresholds of audibility for calibrated speech sounds. If the aided threshold for the /s/ speech sound is elevated, the audiologist may consider providing stronger frequency-lowering settings. Also, UWO has developed the Plurals test, which presents simple words in the singular and plural form (with /s/ in the final position – e.g., cat, cats, dog, dogs). The child's task is to determine whether the test word contained an /s/ in the final position to indicate plurality. The audiologist should also maintain a constant dialogue with the child's listening and spoken language specialist (LSLS) and caregivers to get updates on the progress and/or concerns regarding the child's receptive and expressive communication skills and ability to hear and produce high-frequency speech sounds.
3. Will my patients immediately show benefit when I fit frequency-lowering technology, or will it take a while?
In our experience, many children show immediate benefit from the use of frequency-lowering technology, particularly if they do not have audibility for acoustic information above 4,000 Hz without the use of frequency-lowering processing. In research studies and clinical sessions, we have seen children who miss every /s/ in the final position on the UWO Plurals test without the use of frequency-lowering technology but then immediately score near 100 percent correct on the Plurals test when frequency-lowering processing is enabled. However, it should be noted that we and other research teams have observed improvement in performance across time as the listener acclimates to frequency-lowering processing. For example, in a study of the benefit of frequency-lowering technology for children with moderate hearing loss, Wolfe and colleagues reported better speech recognition in quiet and in noise after six weeks of use of non-linear frequency compression (NLFC) relative to their performance measured on the day when NLFC was enabled (Int J Audiol. 2011). Glista and colleagues also showed improvements in speech recognition as children acclimated to NLFC, but benefit typically stabilized by the two-month post-fitting point (J Speech Lang Hear Res. 2012 Dec;55(6):1765). In short, individuals may require as little as a few minutes to as many as several weeks to adapt to frequency-lowering technology, but acclimatization is usually complete after two months of experience.
2. Are there any differences in the frequency-lowering processing technologies offered by various hearing aid companies?
Substantial differences exist in the frequency-lowering technologies offered by major hearing aid manufacturers. Phonak, ReSound, Siemens, and Unitron all use a variant of frequency compression processing, whereas Oticon, Starkey, and Widex all offer frequency-lowering processing that replicates the dominant signal in the high-frequency portion of the spectrum and moves it to a lower frequency range (e.g., linear frequency transposition, frequency composition, frequency translation, etc.). Regardless of whether frequency compression or some other form of lowering is used, there are differences in the way each manufacturer implements its own proprietary type of frequency-lowering processing.
Additionally, some manufacturers offer static frequency-lowering processing that lowers all high-frequency sounds regardless of the composition of the input signal, whereas other manufacturers offer adaptive frequency-lowering processing, in which the amount of lowering differs based on the characteristics of the input signal. Adaptive frequency-lowering processing has been designed to limit the amount of distortion created by the frequency-lowering and to avoid alteration of lower-frequency speech sounds (e.g., the second formant frequency of vowels).
Audiologists should “take a look under the hood” to gain an understanding of the details of the frequency-lowering algorithms employed in hearing aids fitted for children. The differences in the frequency-lowering processing schemes offered by the various manufacturers can influence how these technologies are fitted. For instance, the crossover frequency (e.g., start frequency at which frequency-lowering processing begins) typically must be well within the range of the listener's usable hearing so that the compressed sound is lowered to an area in which the sound is audible and the listener's hearing acuity is capable of resolving the sound. In contrast, the start frequency of linear frequency transposition should be positioned at or just beyond the upper limit of the listener's usable range of hearing because the high-frequency sounds beyond the start frequency are transposed to a lower frequency where audibility may be restored. To appropriately fit frequency-lowering technology, the audiologist must have a clear understanding of the function of the algorithm underlying the processing.
1. Know When to Fold ‘Em!
In another classic country and western song, “The Gambler,” Kenny Rogers reminds gamblers across the country that they have to know when to fold ‘em when they're dealt a bad poker hand. Similarly, audiologists must know when to recognize that the use of frequency-lowering technology will not sufficiently meet the needs of children with severe to profound hearing loss. Figure 6 shows the audiogram of an 8-year-old child who recently experienced a significant shift in his hearing loss and who was subsequently fitted with hearing aids with frequency-lowering technology. A substantial amount of frequency-lowering processing was used to provide audibility of the /s/ speech sound (Fig. 7). As a result, a great deal of overlap exists between the /s/ and /∫/ speech sounds. The frequency-lowering applied in this case is likely to introduce distortion that will hinder speech recognition and sound quality. Given the severity of the child's hearing loss in the mid- to high-frequency range, he should be considered for a cochlear implant in at least one ear.Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.