Valid clinical verification of digital signal processing (DSP) functions is essential for answering the fundamental questions of both clinicians and patients. For the clinician, the most fundamental questions to answer are, “Does it work? Do the unique signal processing functions of this digital hearing aid (ie, its directionality, noise reduction, feedback suppression, etc.) actually do what the manufacturer has indicated they are supposed to do?” In order to answer this question, the clinician should conduct scientifically valid measurements of the aid— both in the test box and on the patient’s ear—to verify that its digital functions are operating as expected. And, since these measurements must be made in a busy clinical practice, they should by necessity be clinically expedient measures.

For the patient, the most fundamental questions to answer are, “Is it Valuable? Does this digital hearing aid system provide me with capabilities and functions that I cannot get any other way, and are these capabilities and functions worth what I am being asked to pay for them?” In order to answer this question, the patient should acquire a meaningful understanding of the new capabilities and advantages being provided by the hearing aid. As is true in any process of understanding, visual and auditory demonstrations of these capabilities will increase the likelihood of their meaningfulness and value. If these auditory and visual demonstrations are scientifically founded, they move the patient’s value equation from subjective judgment to objective awareness.

Part 1 of this two-part article (see last month’s HR)1 explained how to properly attain and verify a fitting that considers both the “audibility window” and the recruitment characteristics of the patient using the Verifit system developed by Audioscan. In Part 2, we will examine how to verify the enhanced directionality, noise reduction, and feedback control features of a digital hearing aid using this same system.

Verifying Directionality in Digital Aids
Directional microphone technology has experienced a popular resurgence in large measure due to enhancements in directional functionality that digital signal processing has introduced. Using input signals from one or more forward-located microphones and one or more rear-located microphones, digital algorithms have been designed to exploit the time and intensity differences in incoming signals. By selectively amplifying sounds originating from a preferred direction, while attenuating those from other directions, digital hearing aids can produce predictable directional microphone sensitivity patterns. Through DSP, the location of directional sensitivity null points and the direction and magnitude of directional lobes can be adaptively optimized in response to changes in listening situations.

The two most common methods manufacturers use to quantify directionality are by generating a polar plot or calculating a directivity index (DI). The polar plot is typically generated by rotating the hearing aid (on a manikin) in front of a pure-tone source in an anechoic chamber. The DI is typically calculated from pure-tone frequency response curves from many sources surrounding the hearing aid (on a manikin) in an anechoic chamber. Neither method is a viable testing option at the clinical level. It is not uncommon for directional microphones to be reverse wired, mis-programmed, or for their ports to be partially clogged. Therefore, it is essential that the clinician have a means to test the functionality of the directional instrument at the clinical level on a routine basis.

f03a.gif (35281 bytes)Figure 1. Dual-speaker test chamber. With the hearing aid oriented at the test point as indicated, the left (L) speaker becomes the front signal source, and the right (R) speaker becomes the rear signal source for directional microphone testing.

The Verifit system offers the clinician an effective means of objectively verifying directional functionality of both linear and nonlinear hearing instruments through the use of a novel broadband test signal which is presented simultaneously through front and rear soundfield speakers or through the two speakers in the dual-speaker test chamber (Figure 1). This method (patent pending) generates two noise signals, one from a front-facing speaker, and one from a rear-facing speaker, each having a unique spectrum. The spectra of these two signals are independently controlled in real time by a single reference microphone and, after being processed by the hearing aid, are separated to produce two, simultaneous, real-time, broadband frequency response curves. As in the real world, both signals are processed simultaneously by the hearing aid. Simultaneous test signal presentations from two signal source locations is necessary to ensure that the hearing aid’s compressor(s) do not “adjust” the relative front and back microphone output results due to the differing input signal levels these microphones may deliver to the processor.

figureFigure 2. When testing an omni-directional hearing aid (or a directional aid with its directivity de-activated) using the dual-speaker/dual-input function of the system, the L (front) and R (rear) speaker output responses overlap.

Figure 2 displays the results obtained when measuring a directional hearing aid in its omni-directional mode in the test chamber while presenting the hearing aid simultaneously with these two spectrally-unique noise signals, each generated from separate, 90°-opposed, chamber speakers. The hearing instrument has been oriented in the test chamber so that the front microphone is facing the left chamber speaker and the back microphone is facing the right chamber speaker. In the omni-directional mode, the “L” (left speaker) and “R” (right speaker) output response curves overlap.

figureFigure 3. When testing the directional function of a directional hearing aid using the dual-speaker/dual-input function of the system, the L (front, pink curve) and R (rear, lower pink curve) speaker output responses separate, quantifying the directional effect.

In Figure 3, the same test has been repeated, but the directional function is now activated. The results of this measurement are displayed as pink curves. Note that the frequency response curve labeled “R” is less intense than the frequency response curve labeled “L”. This difference is due to the directional microphone functionality of the measured instrument. Thus, by measuring curve separation in the directional mode, and confirming the absence of curve separation in the omni-directional mode, the clinician can objectively verify the functionality of the hearing instrument’s directional feature. If the directional characteristics are adaptive, this degree of separation will be seen to increase with time, once the directional feature has been activated. If the directional microphone is wired backwards, the L curve will be lower than the R curve.

It is useful to save the initial directionality measurement obtained on a hearing instrument using this test, and then to repeat the test if and when subsequent directional functionality is questioned. Directional microphone performance compromise due to such factors as microphone drift or a plugged microphone port can then be measured clinically by comparing the latest directional microphone curve separation result with the original “benchmark” measurement.

Verifying Noise Reduction in Digital Hearing Instruments
By analyzing the modulation patterns of incoming sound, many digital hearing devices can be programmed to detect modulation patterns of speech. Using such speech modulation detection in each of the independent frequency bands, digital aids can be programmed to reduce amplification in those bands where speech modulation is not detected. The result is that the loudness of non-speech input energy residing in amplification bands that do not contain speech-modulation energy is reduced. The amount of amplification reduction that is introduced using this technique varies from one manufacturer’s design to the next, and can even be adjusted in some aids by the clinician programming the aid. The “attack” and “release” times of this effect also varies by technology design.

figureFigure 4. In the presence of “pink noise” stimuli, the green (reference) curve corresponds to the hearing aid output with the noise reduction effect not yet activated. The pink curve is the hearing aid output once noise reduction is fully engaged.

Using real-time spectral analysis and presenting broadband noise to a digital hearing instrument equipped with a noise reduction feature, it is possible to engage and measure the noise reduction functionality in all bands. Additionally, one can quantify the noise reduction time constants, as well. Figure 4 displays two frequency response curves obtained when subjecting a digital hearing instrument with noise reduction to 70 dB pink noise while coupled in a test chamber. The top curve is the pink noise generated frequency response with the noise reduction feature off. The bottom curve is the pink noise generated frequency response with the noise reduction feature active (after the effect has been fully engaged). The separation between these two curves quantifies the magnitude of the noise reduction effect across bands.

Since the Verifit system computes a new spectrum every 128 milliseconds, it is possible for the patient and the clinician to “watch” the noise reduction feature engage. To do this, the clinician first generates and stores a stabilized response curve from the test hearing instrument with its noise reduction feature off, or not yet activated. This becomes the reference curve. Then, after turning the noise reduction feature on and ensuring that the hearing instrument is in the same test position used for the first recording, a second curve is generated and displayed on the same screen that the reference curve has been stored. While in the active measurement mode, the second curve will initially “evolve” and stabilize over the originally stored curve. Then, over time, the second curve should begin to decrease in level relative to the reference curve. This decrease in curve position will continue until the full noise reduction effect has been activated (ie, the curve stabilizes at a reduced output level). The noise reduction attack time is defined by the amount of time it takes from stimulus onset until the second, lower-level stabilized result has been achieved.

This same measurement condition can be duplicated with the system using a free-field speaker source, and measuring the hearing aid output with a probe microphone while the patient is wearing the aid. This clinical approach allows the patient to both “see” and “hear” the activation of the noise reduction feature while wearing the device. The system’s output monitoring headset (which monitors either the probe microphone or coupler microphone response) also allows a third party (ie, clinician or patient’s companion) to hear the noise reduction feature as it is activating.

Verifying Digital Feedback Suppression Functionality
Most digital hearing instruments are equipped with some measure of feedback reduction technology. Digitally based approaches to feedback containment can be either passive or active. In passive approaches, an attempt is made during the fitting procedure to identify the presence of feedback oscillation and its frequency. Then, gain in the band containing that frequency is reduced sequentially until the oscillation is maximally contained. In active approaches, an ongoing oscillation detection mechanism is built into the digital design. If oscillation (presumed to be feedback) is detected at any time during ongoing hearing aid use, the system activates its feedback reduction function in an effort to suppress that oscillation. Active feedback suppression systems can use either automatic band gain reduction (notch filtering) or digital phase canceling (a digitally created out-of-phase oscillation wave that cancels the audibility of the oscillation).

The value of effective feedback containment is best demonstrated to both the patient and the clinician when the hearing instrument can be operated at the necessary gain and output levels without feedback interference. If feedback was a problem at these levels prior to feedback suppression activation, or during a previous hearing aid use experience, the value of digital feedback suppression will be readily understood and appreciated.

However, the clinician will also want to determine if the hearing instrument’s feedback containment mechanism is introducing any unwanted changes in overall hearing instrument performance that may compromise other aspects of the aid’s utility. For example, if a significant change in the frequency response has been introduced in an effort to contain feedback oscillation, this change may reduce the effectiveness of the hearing instrument to deliver appropriate audibility for other important input signals.

f03e.jpg (10301 bytes)Figure 5. The test configuration used to verify digital feedback suppression includes a coupled hearing aid in an open test box with the monitoring headset placed in close proximity to the coupled hearing aid.

Figure 5 depicts a testing configuration that can be used to verify both the functionality of digital feedback suppression technology and the impact that functionality may have on the overall output performance of the hearing aid. The target hearing aid is placed in the test box, coupled to the appropriate 2cc coupler in the same fashion as would be used for any conventional coupler test procedure. However, in this case, the chamber lid is kept open, and the system’s monitoring headset is placed near the test point.

In the above setup, with the test hearing aid on, but with its feedback reduction technology off, an output measurement is obtained in the presence of 60 dB pink noise. The green curve in Figure 6 depicts this measurement outcome. Next, the above input condition is repeated while monitoring the output result during real-time analysis. Initially, the second measure readily overlaps the first. While monitoring this real-time output measure, the volume control of the monitoring headset is gradually turned up until the hearing aid begins to produce feedback oscillation. This will be readily heard, as well as visually evident by the appearance of an oscillation spike on the measurement curve. This curve is then stored (pink curve in Figure 6). Finally, while the hearing aid is still oscillating in the chamber, the feedback suppression technology of the hearing aid is engaged. Once the feedback suppressor is engaged, a third curve is obtained in the exact same test conditions that were present when the second curve was stored (blue curve).

figureFigure 6. Feedback verification results. The green curve represents the coupler response to 50 dB pink noise stimuli. The pink curve represents coupler response with monitoring headset volume control turned up to create hearing aid oscillation (ie, feedback). The blue curve represents the response obtained with the hearing aid’s feedback manager activated.

By comparing the first and third measures, the clinician is not only able to verify that the feedback reduction technology has eliminated the oscillation, but he/she can also quantify what impact (if any) the oscillation containment process has had on the overall output functionality of the aid.

This same comparison can be obtained on the patient’s ear via real-ear measures. By comparing the output result obtained with the system using average speech stimuli (with the feedback management system active) to the previous result obtained with average speech stimuli (with the feedback management system inactive), any alterations in hearing aid performance for speech signals that the feedback containment circuitry has introduced becomes evident. If such adjustments are deemed unacceptable in delivering desired speech audibility performance, then more conventional feedback containment techniques may need to be used instead of digital feedback control.

Summary
Through the use of recently developed clinical measurement technologies, it is now possible for the clinician to verify and demonstrate key digital hearing aid performance functions in the clinical setting. Four key digital hearing aid performance features—multi-band wide dynamic range compression (WDRC), digitally controlled directionality, digital noise reduction, and digital feedback suppression—can now be objectively measured and their functionalities scientifically verified in the clinical setting.

By measuring real-ear hearing aid output in the presence of soft and average speech input stimuli, and properly positioning the resulting output spectrum within the patient’s output-based “audibility window”), the clinician can clinically verify and demonstrate the functionality and advantages of WDRC and digital filtering (see Part 1 of this article).1

It is also possible to clinically measure the “front” and “back” microphone frequency response curves produced by a directional hearing aid in the presence of simultaneous front and back speaker stimulation. Separation of these curves when the directional feature is on (in comparison with no separation of these curves when the directional feature is off) verifies directional functionality. This test can also be used to detect microphone drift or a plugged microphone port by comparing the current directionality test results to the original “benchmark” measures obtained at the initial fitting.

Digital noise reduction technology can be objectively and quantifiably measured and demonstrated using the real-time data-gathering technology and pink noise stimuli. The active output measurement of noise reduction in multi-curve mode will quantify the noise reduction function over time by displaying a gradual reduction in curve position on the display screen as the noise reduction feature engages. This output reduction can also be audibly monitored through the measurement microphone monitoring headset, providing both the clinician (or another person) an audio demonstration of the noise reduction feature’s effects.

Digital feedback suppression technology can be most effectively verified and demonstrated by eliminating feedback interference through its activation. However, the clinician must also confirm that activating the feedback suppression function does not compromise the audibility results previously established with speech-based output real ear measures. Measuring speech-based output results with feedback suppression technology activated, and comparing these results to the same measures without feedback suppression activated, will document any frequency response effects feedback suppression may have introduced.

By being able to scientifically measure, quantify, and demonstrate these four digital hearing aid performance properties in the clinic, the clinician can confirm their functionality, and can more effectively demonstrate their utility to the patient with both visual and auditory demonstration tools. This scientifically based verification and the corresponding increased understanding of the value of digital processing should greatly enhance digital hearing aid acceptance.

Reference
1. Smriga D. How to measure and demonstrate four key digital hearing aid performance features. Part 1: The “audibility window” and precise recruitment accommodation. Hearing Review. 2004;11(11):30-38.

This article was submitted to HR by David J. Smriga, MA, a clinical audiologist, and the founder and president of AuDNet, Burnsville, Minn. Smriga also serves as a consultant for AudioScan, a division of Etymonic Design, Dorchester, Ontario. Correspondence can be addressed to HR or David Smriga, AuDNet, PO Box 1995, Burnsville, MN 55377; email: [email protected].