This blog is a continuation of last week’s blog that focused on hearing aid acoustic feedback. That blog provided general information about feedback and ended with a discussion about acoustic feedback solutions in hearing aids. The intent was to review solutions to feedback that have occurred over the years in somewhat of a chronological order.
The first two solutions presented last week related to 1) overall gain reduction and 2) high-frequency amplification reduction. Both of these were basic approaches and related to early techniques to manage hearing aid feedback solutions, even though some of the current approaches do not exclude these necessarily. For example, limiting the overall sound pressure level continues to be employed, and as such, often prevents hearing aids from reaching oscillation levels. And, although not being discussed in this presentation of electrical approaches to managing acoustic feedback, mechanical means, such as the use of filters in the hearing aid plumbing, also assists in smoothing the frequency response curve – reducing peaks or reducing overall levels, both of which allow for greater stable gain before feedback.
Figures 4 and 5 review by illustration the first two methods of feedback reduction discussed in last week’s blog.
Acoustic Feedback Solution #3 – Electronic Damping of High Frequency Peaks
The idea of using a time-varying system, such as pitch shifting, delay modulation, or frequency shifting, to increase the maximum gain of a system before oscillation occurs, dates back many decades. In 1962, Manfred Schroeder (of digital reverberation fame) published an article in the Audio Engineering Society Journal about using frequency shifting as a method of increasing the gain of a PA system by up to 6 dB. These early PA system attempts to manage feedback are worthy of discussion because some of the hearing aid feedback reduction systems are modifications of these.
Electronic damping of high frequency peaks as a feedback approach engages the electrical circuit of the hearing aid to shift the primary peak of the receiver to a lower frequency to reduce its amplitude and thus, limit acoustic feedback. This is accomplished by either the dispenser or patient varying an adjustable tone control potentiometer on the hearing aid (low-pass filter in Figure 6). Because this process also lowers the peak energy, it has the effect of also damping high frequency peaks. These peak-damping approaches might be thought of as first-order frequency shift approaches, especially since they did not involve multiple channels or bands in a hearing aid, suggesting that fine control was limited.
While the approaches by different developers for frequency shift with its resulting damping are not exactly the same, their general intent and results are – to limit or smooth the peaks of the response in the frequency region where feedback is most likely to occur. As can be seen in Figure 6, this lowering of the frequency peaks is essentially another way to reduce high-frequency gain to reduce feedback. Figure 6 shows two extreme conditions between which the frequency response can be tailored by adjusting the resonant peak control. Adjusting the resonant peak control of the hearing aid varies its electrical output to tailor the frequency response to shift and flatten the peaks, which reduces feedback with minimal reductions in high frequency amplification and only a minimal change in the saturation curve (Figure 7). In many ways, the end result is not too dissimilar from Feedback Solution #2 of this series that lowers high-frequency gain to reduce feedback. In actual use, the slight reduction in high frequency amplification is most likely not detected by the patient because the improved flatness of the frequency response curve (B) permits a higher gain to be utilized. The way these systems are actually fitted is for the patient to adjust the hearing aid to a most comfortable level (MCL), and if feedback occurs prior to reaching the MCL, the control is adjusted until feedback is eliminated.
Acoustic Feedback Solution #4 – Bandpass Filtering
As was mentioned previously, early approaches to manage acoustic feedback in hearing aids can be traced to methods controlling feedback in public address (PA) systems. The delay in implementing many of these feedback management systems in hearing aids was due to the small size and limited space available in hearing aids. The big difference between the PA systems and hearing aids, however, is that most PA systems manage feedback in a fairly static environment, whereas hearing aid feedback is due to any number of ongoing and changing environments and cannot be managed by a feedback system based on a single, primary set of feedback conditions. This has made it more difficult to control acoustic feedback in hearing aids.
Admiraal and Cardozo, working for Philips in the Netherlands (1979)  used high-pass filtering as a way to separate a part of an amplified signal and use that as a control signal by which the gain of the system is reduced (Figure 8). The object was to suppress spurious signals produced as a result of acoustic feedback using a fast control signal. The high-pass filter output was connected to a periodicity detector that discriminated between signals with a high periodicity and those with an irregular character, such as noise or speech – the control signal derived from the output of the periodicity detector. The amplification system would act so fast (less than 0.1 sec) that any further buildup of an acoustic feedback wave was prevented. Additionally, the system would only respond if there was a tendency toward feedback and did not reduce feedback due to treble tones.
Arcos et al (1995)  proposed a feedback reduction method for hearing aids in which the circuit measures short-term and long-term energy in each of eleven bands. Gain is reduced whenever any band sensed a steady-state high-energy signal (assumed to be oscillation from acoustic feedback) within that band. It is not know if this found its way into a wearable hearing aid from the laboratory.
When digital control of analog hearing aids came into existence in the late 1980s, hearing aids having multiple bands (channels) *of amplification became commonplace. The first were primarily 3-channel hearing aids, which were essentially a three-bandpass-filter programmable hearing aid. These instruments allowed for parameter adjustment in one channel to reduce the gain at and around the feedback frequency without substantially affecting the gain in adjacent channels. The major drawback was that with smaller numbers of channels, a fairly large frequency response range was affected. It stands to reason that a greater number of channels would result in more precise adjustments to be made.
* For this blog, the term channels will be used even though it is clear that at the time of these feedback approaches, these were primarily bands and not channels in the true sense of the word.