By Jennifer Groth
Back in the day, audiologists relied heavily on acoustic means to adjust the frequency response of hearing aids. As programmable, multi-channel devices became available and then commonplace, the convenience of adjusting the gain-frequency response through electronic means reduced the use of acoustic modifications to hearing aids to enhance or dampen certain frequency areas. These days, hearing instruments can be fit and fine-tuned using a dizzying number of frequency controls – up to 20 in some products. ReSound hearing instruments can be adjusted via up to 9 discrete frequency gain handles at audiometric octave and interoctave frequencies from 250 Hz to 6 kHz.
We are often asked why ReSound does not offer one frequency gain handle for each of the 17 bands in the Warp Compressor. The short answer is that 9 gain handles provides sufficient resolution to match targets within the bandwidth of the device, but is still easy for the fitter to manage, and can be presented rather simply in the fitting software. Aazh and Moore (2007) reported on the accuracy of real ear match to targets for four hearing instruments, one of which was the ReSound Pixel with only 7 gain handles. They fit a wide variety of hearing losses both occluded and open, and they selected the manufacturer’s NAL-NL1 implementation and verified against this prescription in the real ear equipment. They found that the match to targets “out of the box” was 62% for Pixel, and less than 30% for the other devices tested. After fine-tuning, it was possible to match targets in 100% of the cases with Pixel, and only 60 to 80% for the other devices. Some of this difference was attributable to high frequency gain limitations due to inferior feedback cancellation, as well as each manufacturer’s implementation of the fitting rule, but the fine-tuned results in particular were also explained by the fewer gain handles available to the fitter in the other products. The authors concluded that more gain handles increased the chance that prescriptive targets could be met, although they did not specify whether more than 7 would be advantageous.
Short of repeating a study on the scale of Aazh and Moore, we were interested in finding out whether the resolution available in products with many gain handles really matches the actual performance in the device. In other words, is what you see represented on the fitting screen really what you get in the hearing instrument? To test this, we programmed hearing instruments from five manufacturers to the default first fit for a mild-to-moderate hearing loss. We turned off special signal processing in order to examine only the effects of changing the gains. Then we measured the gain for a 65 dB pink noise input in a 2cc coupler on the Verifit as we made adjustments to the gains in the fitting software. For the sake of simplicity, we present here our findings for gain changes at 1 kHz (or the frequency closest to 1 kHz) and the next consecutive frequency, which was either 1.25 kHz, 1.3 kHz, 1.5 kHz or 1.6 kHz depending on the manufacturer. The baseline frequency response was normalized to 0 dB for each manufacturer, so that the changes are directly comparable in the graphs.
1. What happens when you increase one frequency handle by 1 dB?
Four of the five hearing instruments showed a 1 dB change at 1 kHz as expected. Product B with 20 handles seemed to have a slightly more specific effect at the discrete frequency. Product C with 16 handles showed a much larger change that was broader in frequency than what was shown in the fitting software.
2) What happens if you adjust one frequency a little bit up and the next one a little bit down?
As above, four of the hearing instruments showed changes that corresponded fairly well to what was displayed in the fitting software. Products A, B and the ReSound Alera were the most accurate. Again, Product C was measured to have much greater gain changes than the 1 dB adjustments made in the software and shown on the fitting screen.
3) What happens if you increase one frequency handle by 5 dB?
As the magnitude of the gain change increases, we would expect a very targeted change for a system with steep filters and a broader effect for a system with more overlapping filters. The ReSound system has a large degree of overlap between the bands, which is one of the reasons for not providing 17 frequency gain handles. A change in one band affects other bands, and thus it makes less sense to give individual control over them. In these measurements, we saw a clearer difference between Product B and the others in that Product B does indeed make a change that is quite frequency specific. Product A, Product D and the ReSound Alera showed a broader effect, indicating more overlap in each system’s frequency bands. Product C again showed a huge change compared to what was asked for in the fitting software. The ReSound Alera was most accurate in terms of the 5 dB change at 1 kHz in that the change shown in fitting software corresponded precisely to what was measured in the coupler.
So is what you see in the fitting software really what you get? Our measurements indicate that a greater number of gain handles can provide more discrete frequency control as shown with Product D, but that this is not a given as it depends on the resolution of the underlying system. As an aside, it should also be mentioned that overlapping filter structure can provide sound quality benefits that are ultimately more important to the hearing instrument user than very fine frequency response control. Finally, the actual dB change may be less precise than what is shown in fitting software. Product C is the most chilling example of this, as the device consistently provided more gain than the changes show in software. The ReSound hearing instrument exhibited the most accurate match between the changes made in the software and the measured changes. In addition, the effect on neighboring frequency regions was also accurately depicted.
References
Aazh H, Moore BCB. (2007) The Value of Routine Real Ear Measurement of the Gain of Digital Hearing Aids. J Acad Am Audiol 18:653-664.