Device and Fitting Protocol for a Transitional Intervention for Debilitating Hyperacusis

Abstract

Purpose:

This report describes a hearing device and corresponding fitting protocol designed for use in a transitional intervention for debilitating loudness-based hyperacusis.

Method:

The intervention goal is to transition patients with hyperacusis from their typical counterproductive sound avoidance behaviors (i.e., sound attenuation and limited exposure to healthy low-level sounds) into beneficial sound therapy treatment that can expand their dynamic range to the point where they can tolerate everyday sounds and experience an improved quality of life. This requires a combination of counseling and sound therapy, the latter of which is provided via the hearing device technology, signal processing, and precision fitting approach described in this report. The device combines a miniature behind-the-ear sound processor and a custom earpiece designed to maximize the attenuation of external sounds. Output-limiting loudness suppression is used to restrict exposure to offending high-level sounds while unity gain amplification maximizes exposure to healthy and tolerable lower level sounds. The fitting process includes measurement of the real-ear unaided response, the real-ear measurement (REM) system noise floor, the real-ear occluded response, real-ear insertion gain, and the output limit. With these measurements, the device can achieve the prescribed unity gain needed to provide transparent access to comfortable sound levels. It also supports individualized configuration of the therapeutic noise from an on-board sound generator and adaptive output limiting based on treatment-induced increases in dynamic range.

Results and Conclusion:

The utility of this device and fitting protocol, in combination with structured counseling, is highlighted in the outcomes of a successful 6-month trial of the transitional intervention described in a companion report in this issue.

This report describes a novel device and fitting protocol, which, together with an accompanying counseling protocol detailed in this issue (Cherri et al., 2024), is the basis for a patented transitional intervention (Eddins et al., 2020) for treating debilitating primary loudness-based hyperacusis (LH; Tyler et al., 2014). Companion reports in this issue review the background and rationale for the transitional intervention (Formby, Secor, et al., 2024) and the outcomes of a successful 6-month field trial of the intervention (Formby, Cherri, et al., 2024). Below, the device technology used in the transitional intervention is described, along with the step-by-step protocols for fitting, setting, and adjusting the treatment-driven adaptive loudness suppression (LS) used to limit exposure to offending sound levels. The nature of the therapeutic sound used for recalibrating hypergain neural activity within the central auditory pathways (the presumptive mechanism for LH) is described along with the procedures for establishing individual sound therapy levels. Optional configurations for use of the sound generators with and without sound protection are presented.

A Transitional Intervention Device for LH

Many individuals with hyperacusis are habituated to isolating conditions that restrict their exposure to offending sound levels. Thus, it is challenging for them to transition from their typical very quiet environment to a therapeutic sound-enriched environment and ultimately to routine sound environments and commonly experienced public places. The goals and methods of this intervention are delineated in Figure 1.

Figure 1.

Transitional intervention for hyperacusis: goals and method.

Briefly, we seek to achieve five main goals. The first goal is to prepare the affected individual to transition from the safety of their self-imposed sound avoidance to the starting point for their treatment. This point of treatment initiation is marked by the willingness to accept a protected yet sound-enriched environment and the use of therapeutic sound. This initial step is promoted in a companion-structured counseling protocol detailed by Cherri et al. (2024).

The next four goals are achieved primarily with the use of our protective treatment devices, which we detail below in this report. This includes the second goal, which is to provide controlled therapeutic sound. This neutral, low-level, broadband “seashell”-like sound drives the treatment-induced changes in our transitional intervention that lead to resolution of LH. The therapeutic sound is produced by sound generators onboard our bilateral ear-worn devices. The controlled therapeutic sound typically is maintained at a “soft, but comfortable” level over the course of the intervention to promote recalibration of the “hypergain” processes that give rise to LH. It is this recalibration process that leads to acceptance of sounds previously considered to be uncomfortably loud and ultimately to the resolution of LH. The controlled active sound therapy achieved with the ear-worn devices is supplemented in our approach by safe and healthy passive exposure to neutral environmental sound sources within the home.

The third goal is to provide the desired protection from sound levels that are uncomfortable to the wearer. This is accomplished in two ways. First, each ear-worn device includes a custom earpiece designed to function as a very effective protective earplug when used in isolation (e.g., connected device is muted or turned off). When the earpiece is connected to the active behind-the-ear (BTE) intervention device, amplification is provided to achieve unity gain; the resulting amplification offsets the insertion loss from wearing the attenuating earpiece. Consequently, transparent hearing (as if no earpiece is worn at all) for low- and mid-level sounds, which are judged by the wearer to be acceptable, is attained. This feature, in turn, makes possible the second level of protection achieved with output-limiting LS, which restricts exposure to potentially offending high-level sounds.

The fourth goal is to wean the LH sufferer from their misuse of hearing protection, including routine overuse of ear plugs and earmuffs, knowing that it is almost impossible for this population to stop abruptly their use of hearing protection devices (HPDs). This is accomplished, in part, by providing the passive attenuation from the earpiece and the active protection achieved with LS via output limiting.

The fifth goal is to reduce sound avoidance behaviors and the use of HPDs, leading to increased exposure to safe and healthy environmental sounds. This goal is promoted by ongoing use of therapeutic sound facilitated by gradual increases in the outputs of the hearing devices monthly (or as needed) based on treatment-driven patient judgments of “loud, but OK” running speech. As sound tolerance improves with treatment, we expect the corresponding loudness levels to increase over the course of the intervention and the associated device output levels to increase accordingly.

Thus, as a consequence of the transitional treatment, the patient with LH is progressively exposed to higher levels of safe and acceptable environmental sounds over the course of the intervention as their sound tolerance and hyperacusis condition improves. In turn, need for both sound protection, including LS, and sound therapy from the sound generators is reduced progressively over the course of the intervention.

Sound Therapy

Individualized implementation of therapeutic sound is a primary component of our comprehensive management and intervention protocol for treating and resolving LH. The treatment motivation for sound therapy with bilateral sound generators is straightforward. The underlying assumption is that the mechanism that gives rise to LH is maladaptive (abnormally increased) neural activity within the central auditory pathways. This hypergain process can be recalibrated (i.e., down-regulated) and reset through use of controlled healthy sound exposure with the goals of reestablishing normal perception of loudness and typical sound tolerance. The premise is that safe and controlled exposure to ongoing neutral sound serves as enriched stimulation and input to the central auditory pathways, thus augmenting and restoring the typical background sound input that has been unavailable to the hyperacusic sufferer in protected isolation. It is this controlled enriched sound stimulation that initiates the adaptive incremental process for recalibration of the maladaptive neural gain.

Our sound therapy protocol is implemented in a two-prong approach. First, in controlled sound environments in which unexpected exposures to high-level sounds were unlikely (e.g., at home in quiet conditions), the use of an open-dome configuration, providing exposure to active low-level sound from bilateral sound generators, is encouraged. This therapeutic sound from the sound generators supplements enriched passive exposure to environmental sound via the unobstructed direct sound path from the environment to the ear. Second, in uncontrollable sound environments or those with high-risk for offending sound levels, the device can be coupled to the custom occluding earmold, which affords protection from offending sound levels via LS while continuing to deliver the therapeutic sound via the onboard sound generator of the BTE device. Both treatment prongs emphasize avoidance of silence at all times.

As successful treatment progresses, and the central auditory gain processes are recalibrated by the therapeutic sound, evidence of these treatment-related effects can be and, indeed, have been observed objectively and subjectively in improved sound tolerance measures, questionnaire scores, and self-reports (see Formby, Cherri, et al., 2024). In turn, as sound tolerance improves, this improvement allows for the output-limiting compression threshold set in the protective devices to be increased systematically. Subsequently, higher sound levels, which now are comfortable for the patient, can be presented while continuing to suppress the levels of offending sounds. The result of this iterative treatment process is reduced need for output limiting, greater exposure to sound levels considered acceptable by nonhyperacusic individuals, and expansion of the dynamic range for hearing. Thus, with the combination of earplugging, unity gain, healthy exposure to comfortable sounds, positive sound therapy treatment, and targeted counseling (i.e., Cherri et al., 2024), the hyperacusic sufferer can be incrementally and comfortably exposed to an extended range of safe and healthy sound levels that, ultimately, are necessary for attaining and sustaining typical sound tolerance.

Protective Hyperacusis Management

In concert with sound therapy, protection from offending loud sounds using active LS via output-limiting compression is supported by the same custom-fit devices. As detailed by Formby, Secor, et al. (2024), this approach follows from Sammeth et al. (2000), who described and evaluated one of the early applications of an active LS strategy for protective management of LH. Their goal for LS was to replace the detrimental sound deprivation effects of earplugs and earmuffs with instruments featuring protective (infinitely) compressive output limiting. Their approach required the use of a fully occluding earmold whose sound-attenuating properties were offset by unity gain amplification. The resulting LS instrument was effectively an infinitely compressing analog hearing aid (HA) that limited the listener's exposure to uncomfortably loud sound levels while preserving typical hearing for soft and comfortable sound levels without compromising audibility. Although not formally evaluated in a prospective study, their hyperacusic participants, when fitted with the LS instrument, achieved wider dynamic ranges compared with those measured when sound-attenuating earplugs were used. In some cases, measurable improvements in speech understanding and enhanced subjective sound tolerance were reported (Sammeth et al., 2000). Unfortunately, their innovative approach, while promising, was not developed or promoted further for clinical use, perhaps because the technology available at that time was not sufficiently advanced for the challenges and ease of customized fitting (i.e., setting unity gain) for a broad range of individuals with severe hyperacusis. Moreover, the LS approach described by Sammeth et al. (2000) was one of enhanced protective management (better than a sound-attenuating earplug), rather than a treatment, as they considered that LS might be useful when combined with an intervention such as tinnitus retraining therapy (Jastreboff & Hazell, 2004). They offered no companion counseling component in their evaluation of LS, nor did they describe iterative adjustment of the LS settings over the course of their evaluation period. However, as we report in the Formby, Cherri, et al. (2024) study, the principles of their protective LS approach have significant practical and clinical utility. Below, we describe how we have extended and enhanced the setting and implementation of LS with today's modern technology as a tool in our transitional intervention for treatment of debilitating LH.

BTE Component

The protective therapeutic device was designed and patented (Eddins et al., 2020) by the co-authors, Eddins, Formby, and Armstrong, and assembled by General Hearing Instruments (GHI), now SoftTouch Labs. Its constituent components and fully assembled forms are shown in Figure 2. Device components are housed in a miniature BTE shell. Components feature two microelectromechanical systems (MEMS) microphones (Knowles, MQM-33010-000); a low-noise, moderate power receiver (Knowles Electronics, RAB Series); a size 312 battery supply; and, at the core, a digital signal processing (DSP) chipset manufactured by On-Semiconductor (Model 3910). The MEMS microphones are small microphones with very high fidelity, low noise, stable frequency response, and high signal-to-noise ratio. They are mounted in the body of the BTE shell in a typical front/rear configuration. The receiver is mounted forward in the case and mated to a standard thin tube (Intech, Model 00A-1098503101). The length and bends of the thin tubes are customized to fit into the left and right ear canals of each user.

Figure 2.

Images of the therapeutic device designed in-house and assembled and tested by General Hearing Instruments (now SoftTouch Labs). Left: Hearing device with thin tube and attached open dome earpiece. Center: Hearing device with custom earpiece containing a heat-activated stent as used in the field. Right: Custom earpiece containing the heat-activated stent and a probe tube embedded in the earpiece to support precise acoustic measurements during the fitting and adjustment processes. Embedding the probe tube minimizes probe tube–related slit leaks that occur when the probe tube is placed between the earmold and the canal wall.

In-the-Ear Components

In one configuration, the thin tube is connected to the BTE device implemented with an open-dome receiver (sizes small, medium, or large, depending on user needs). The latter is positioned in the ear canal and serves as a maximally vented retention piece for transparent transduction of environmental sound directly into the ear canal in conjunction with presentation of the therapeutic sound. In a second configuration, the BTE device connects to a custom-occluding earpiece via thin tube, also supplied by GHI. In our field trial, described in the companion results paper (Formby, Cherri, et al., 2024), each earpiece was based on accurate ear impressions made by the research audiologist using Audiologist's Choice DS-50 (medium viscosity; Oaktree Products) impression material (see Appendix A). The custom earpieces are constructed from medical grade silicone (Factor II; Lakeside Arizona) and contain a heat-activated stent (Juneau et al., 2010) that expands in the ear based on change from room to body temperature. The precise physical fit of the custom earmold within the ear canal, combined with its soft, pliable silicone structure and the heat-activated stent, result in an earpiece that functions as a high-quality earplug when there is no sound input from the device. The sound-attenuating effects of the earplug vary by frequency and range from 25 to 30 dB at the low frequencies (250–1000 Hz) up to 40 dB or more at high frequencies (4000–8000 Hz). In addition to the open-dome configuration and the custom-occluded earpiece configuration, an identical custom earpiece is created that includes a permanently embedded silicone probe tube that passed through the full length of the earpiece. This third earmold configuration facilitates acoustic measurements of the sound pressure level (dB SPL) in the ear canal during initial device fitting and subsequent adjustment (as described in the device fitting protocol below). When this earpiece is placed within the ear canal, the opening of the probe tube is positioned approximately 5 mm from the tympanic membrane.

Device Function

The intended function of the therapeutic device mirrors several properties of fully occluding HA devices, and device fitting involves several adaptations of standard validation methods. A major difference is that the tolerances for each function are more stringent than for typical hearing devices; thus, validation requires greater precision and careful control of sound level at all times to avoid overstimulation of the wearer with hyperacusis. The major functions and corresponding verifications are illustrated in the schematic diagram of Figure 3. A first step is to measure the individualized, ear-specific, real-ear unaided response (REUR) as a function of frequency (i.e., with no device inserted). This is typically displayed as an output–frequency response curve. In Figure 3, however, it is schematized for a single frequency on an input–output diagram as the leftmost thin dashed line. The occluding earpiece function, when inserted, is captured by the real-ear occluded response (REOR) as shown by the rightmost thin dashed line for the same frequency. As shown, the REUR and REOR are linear with changes in input level. Importantly, the vertical offset between the two functions represents the passive attenuation of the earpiece at that frequency. With the occluding earpiece in place and the device amplification active, the sound picked up by the microphones is amplified, and unity gain can be achieved. In this case, the real-ear aided response (REAR) matches the REUR of the individual subject (leftmost thick dashed line) for input sound levels below the output-limiting level, and the earplugging effect of the earpiece is fully compensated and offset by the unity gain. The resulting device output preserves the sound level in the natural environment and achieves the goal of providing healthy exposure to comfortable sounds at lower sound levels.

Figure 3.

A series of input/output functions illustrating the function of the device (see descriptions in the text) at a single (unspecified) frequency. The REUR for that frequency is shown as a function of level by the leftmost green line. The REOR obtained with the fully occluding earpiece is represented by the rightmost green line. The vertical difference between REOR and REUR represents the passive attenuation produced by the earpiece. With unity gain, thick dashed black line in the lower left shows that the REUR is restored for input levels below the output limit. The output limit is based on “loud but OK” judgments for running speech at the baseline established at Visit 0. Above this level, the REOR is preserved, and corresponding loudness is suppressed for sound levels above this limit. The output limit is progressively increased with treatment success until the final visit. The difference in output limit from baseline to final visit is the treatment effect. REUR = real-ear unaided response; REOR = real-ear occluded response; REAR = real-ear aided response.

With output limiting, however, the REAR can match the REOR (rightmost thick dashed line) for levels above the output limit setting, and the loudness perception of those high input levels is suppressed. The output-limiting feature follows the LS strategy used by Sammeth et al. (2000). LS allows high-level, potentially offending sounds to be attenuated by as much as the full effect of the earmold attenuation. Labeled as “baseline” in Figure 3, this output-limiting effect ensures that patients can engage comfortably in activities in a variety of acoustic environments, some of which may otherwise place them at risk for exposure to offending sounds. Consequently, the patient can enjoy full access to comfortable sound levels without any attenuation at all while having safety and protection from high-level sounds through the LS function.

The LS function is implemented as output amplitude compression with a variable compression threshold (dB SPL) based on patient-specific loudness judgments for continuous speech (see device fitting protocol). The thick dashed line of Figure 3 represents our typical device function with low-level inputs (leftward on the x-axis) matching the REUR and high-level inputs (rightward on the x-axis) matching the REOR. The LS features an eight-channel limiting scheme using infinite:1 compression in each channel. To maintain excellent sound quality, dual-level detector technology is employed. The first-level detector is in control the majority of the time, having a design goal of moderately slow compression speed (32 ms attack time constant, 256 ms release time constant) to preserve the overall sound envelope. The secondary-level detector algorithm, with short (< 1 ms) attack and release times, activates only in response to transient sounds, which might otherwise cause annoyance. With successful treatment, the output limit level can be released over time, as illustrated by the red dashed line labeled Visit 2.

The chipset at the core of the device incorporates a sound generator feature that can be customized by the fitting audiologist. This device is configured to produce continuous low-level, broadband, seashell-like therapeutic sound when used. The level of the sound is set to a “soft-but-comfortable” level as judged by the patient, and the physical level of that sound is documented using the probe microphone measures. In the custom-occluded earmold fitting, the spectrum of the therapeutic sound, measured near the tympanic membrane, has a high-frequency emphasis. This spectrum mimics the spectrum of the broadband sound therapy provided by Formby et al. (2015). In the open dome fitting, the low-frequency portion of the spectrum of the therapeutic sound is reduced due to sound leaking out of the ear canal (as it would with any large vent in an earmold). The minor differences in the spectrum of the therapeutic sound, shown in Figure 4, are not known to compromise the therapeutic effect of the sound generator and have little effect on the perceived level of the therapeutic sound.¹ While it may be possible to compensate for the vent-related changes in spectral shape of the therapeutic noise, by manually switching between two different device programs, the high probability of incorrectly switching between optional programs led us to use a single program knowing and accepting that the spectrum of the therapeutic sound differs slightly for the two earpieces. Basic and advanced device function was specified by the research team and implemented using the low-level software provided by the chipset manufacturer. A custom high-level software suite was developed and used for device evaluation, fitting, verification, and adjustment.

Figure 4.

Sample in situ magnitude spectra of the therapeutic sound, as recorded in and averaged across four ears under two conditions: occluded earpiece (solid line) and open earpiece (dashed line). Greater differences at low frequencies (< 1800 Hz) reflect the high-pass nature of earpiece venting, as described in the Underlying Acoustic Model section.

While the probed earmold is used only by the audiologist in fitting the device, the open dome and standard occluded earmold are made available for use by the patient at any time. Patients are instructed on their use and encouraged to use the therapeutic sound delivered by the devices as much as possible during their awake hours. To maximize the likelihood of extended device use and facilitate this use in various acoustic environments, the open-dome fitting is designed for acoustic environments deemed by the patient to be safe and low risk for offending sound levels. The occluded earmold fitting is designed for use in acoustic environments where there is risk for encountering offending sound levels. The patient is instructed on the removal and replacement of the thin tubes connecting the BTE device to the open dome or occluded earmold. Relevant here was an early concern that the devices and thin tubes might not be robust enough to withstand regular and repeated removal and replacement. A laboratory experiment therefore was conducted in which experimenters removed and replaced the thin tube from the BTE shell 1,000 times. After the 1,000th cycle, both the BTE shell and the thin tube showed no visible or mechanical wear, and the fitting was as secure as the original placement.

Device Fitting Protocol

The device fitting and adjustment protocol (summarized in Appendix B) was implemented in a comprehensive field trial of a transitional intervention for LH (Formby, Cherri, et al., 2024). This trial and subsequent research using this protocol may lead to further streamlining, including a reduction in the number of steps involved in the device fitting protocol, simplification of the fitting software interface, and introduction of methods to minimize the number of laboratory or clinic visits. Because the overall format of the protocol is grounded in years of clinical research (see Formby, Secor, et al., 2024) and the results of the initial field trial indicate real potential to enhance sound tolerance (see Formby, Cherri, et al., 2024), it is unlikely that the general structure of the protocol or its key elements as described here will change substantially.

Acoustic Model and Preliminary Measurements

To facilitate device fitting and design, maximize measurement precision, support troubleshooting, and generate data needed for modeling, the initial fitting process and protocol development were completed on a KEMAR (Knowles Electronics, Type DB 4004) acoustic test fixture. This effort allowed all design decisions to be made based on standardized data. After the final design decisions were made, the desired performance curves to be measured on individual participants were modeled based on the KEMAR measurements. The results of those measurements are stored within the fitting software and can be displayed alongside actual participant data as desired. A goal was to measure the key acoustic parameters that likely would account for any performance differences between the individual subject and model data based on KEMAR measurements. We then could update the predictive models to reflect the actual variations associated with the specific individual, adjust device parameters to obtain a best fit to desired performance targets, and verify device performance on the individual. While it is known that KEMAR captures the acoustic properties of the median individual, by definition, individual real-ear measures will differ somewhat from KEMAR measures. These differences can impact the accuracy of the acoustic performance of an HA coupled to an individual's ear.

As illustrated in Figure 5, the key elements in the generic model include the amplified sound path (upper left) and the direct sound path (lower left). The summation of these parallel pathways (right) represents the sound pressure developed at the tympanic membrane of an individual wearing an ear-level hearing device, such as an HA. The amplified path includes acoustic input at the position of the microphone inlet, which has been modified by the diffraction and reflections associated with features of the pinna, head, neck, and torso relative to sound propagating near the listener. These alterations are represented in the model as the microphone location effects (MLE). The use of KEMAR during the design process ensured that the MLE factor is accounted for in general. Any differences in this factor at the individual level were quite small and were not considered in the fitting protocol.

Figure 5.

Acoustic model representing the amplified path of sound transmission and the direct path of sound transmission, including microphone location effects (MLE), bidirectional vent effects, and corresponding real-ear measures as described in the Underlying Acoustic Model section. HA = hearing aid; RECD = real-ear-to-coupler difference; REAG = real-ear aided gain; REUG = real-ear unaided gain; invREUG = inverse of REUG; REIG = real-ear insertion gain; REOG = real-ear occluded gain.

The common use of a 2-cc coupler as the acoustic load for an HA during test box measurements underrepresents the output measured using real-ear measurement (REM) techniques on human subjects. The residual ear canal impedance on a subject is more complex than what a model based on a simple volume of air can achieve. At an individual level, the acoustic impedance of the residual ear canal volume, interacting with the ear canal termination impedance created by the tympanic membrane and middle ear structure, will impact the acoustic pressure delivered by the HA. This is due to the relatively high output acoustic impedance of HAs. The real-ear-to-coupler difference (RECD) block in the amplified path accounts for this frequency-specific increase in output and gain. Like the common clinical procedure used to measure the RECD, in which the coupler refers to the standardized 2-cc coupler, we computed the difference relative to the standardized ear simulator found in KEMAR.

It is common to introduce an intentional acoustic vent path to the HA system for the purpose of high-pass filtering the gain frequency model. The presence of a vent impacts the HA output (vent out) as well as the direct sound flowing into the ear canal (vent in). Collectively, these vent effects are tightly coupled since the acoustic mass of the physical vent impacts both the overall load impedance faced by the HA and the transfer function of the direct path. For modeling purposes, the vent in corrections were added to the real-ear unaided gain (REUG), resulting in the overall real-ear occluded gain (REOG) model of the direct path.

It is important to recognize that the total sound pressure at the tympanic membrane is the result of summing the individual partial components resulting from the two parallel sound paths. Power summation was used on the magnitude of each component in this application, with phase effects being ignored. This choice was rationalized since the achieved isolation with the device worn with the occluded ear mold was more than 25 dB, resulting in minimal comb-filtering effects. Similar comb filtering was present when the participant uses the open dome configuration, but these effects were mostly masked by the therapeutic sound itself.

The overall real-ear aided gain (REAG) predicted at the tympanic membrane can be transformed to predict the desired real-ear insertion gain (REIG) by either subtracting the REUG, or as we chose, adding the inverse of the REUG (invREUG). For a fully occluded ear canal, the vent out corrections reduce to a simple 0-dB correction across all frequencies. While the vent in for an ideal “ear plug” device would be minus infinity across frequency, in practice, bone-conduction paths (not modeled), and the material properties of the earmold do not lend themselves to easy physical modeling. As an alternative, the direct measurement of REOG was performed with results being applied within the software. With the exception of predicting the HA 2-cc performance, all blocks shown in the model can be assumed linear in their transfer function. Knowing this allows a variety of signal types with differing spectral shapes to be used during their measurement. This linearity property enables good quality data to be gathered while balancing the impacts of measurement noise floor and subject comfort, ensuring a predictive model with desired accuracy. Model predictions are used to make an initial estimate of the REIG needed to result in a flat insertion gain, and subsequent fine-tuning is described below. KEMAR measurements of REUR, REUG, REAR, REOR, and RECD are made available in the fitting software and can be recalled and viewed by the fitting audiologist during the fitting process by clicking a check box in the fitting software.

Fitting Software Overview

An essential component of the fitting protocol is physical measurement of the acoustic environment in the residual ear canal when the therapeutic device is inserted. We chose to use a commercially available REM system for measuring device function and performance during fitting and adjustment. This decision was consistent with our goal of achieving and making available a clinically useful transitional intervention, including the treatment devices described here. At the time the device software and fitting software were developed, REM systems did not have readily available functionality to support direct, iterative communication between the device fitting software and REM system software. However, rapid and direct communication was highly desirable, if not essential, between the two software packages given the large number of measurements required for our fitting protocol and the precision desired for the initial field trial of the protocol. Thus, we designed the fitting software to interact iteratively with the Audioscan Verifit 2 REM system. Our custom software sends instructions for specific acoustic measurements from the fitting software to the REM system, while retrieving measurement results from the REM system for use in the fitting software. In essence, we created a closed-loop fitting system controlled from the custom device-fitting software for real-time data sharing between systems.

During the fitting process, the experimenter views the primary fitting software (shown in Figure 6) on one video monitor and the REM system screen mirrored from within the sound booth onto a second computer monitor (shown in Figure 7). The left portion of the fitting screen displays a series of buttons that allow the experimenter to proceed stepwise through the fitting protocol. Those buttons also trigger transfer of relevant information between the fitting software and the REM system connected to the host computer via Ethernet connection. By naming convention associated with the REM system, the term “Session” represents a series of up to four measurements that can be made in succession using the REM system. The selection of buttons labeled “Create Session” in the fitting software creates a file with data that specifies the next set of measurements and associated parameters to be used by the REM system. This data file can be read into the REM system using the Audioscan Verifit 2 “Restore Session” function. Selection of buttons labeled “Read Session” in the fitting software read a file that contains data that have been recorded and stored by the REM system. Thus, “Create Session” sends information from the fitting software into the REM system, and “Read Session” pulls information from the REM system into the fitting software. In this case, each “Session” denotes a different measurement type as outlined below. All “Sessions” are completed for a single fitting or adjustment for the devices fit to the two ears.

Figure 6.

Screenshot of the main interface in the custom software used for fitting the device. See text for detailed description. REUR = real-ear unaided response; REAR = real-ear aided response; SG = sound generator.

Figure 7.

Screenshot of one of the Audioscan Verifit 2 windows that is mirrored onto the external video display during device fitting and adjustment. UCL = uncomfortable listening level; REUR = real-ear unaided response; REOR = real-ear occluded response; BTE = behind-the-ear.

By default, all measures are completed on the left ear first, followed by the same steps for the right ear. The detailed description below outlines this process only once. The top center of the fitting screen (see Figure 6) displays hearing device communications buttons. Selection of the “Connect” button establishes a connection between the hearing instruments and the computer via programming interface (Sound Design Technologies, DSP Programmer V3.0). The “Write” button resets the device RAM to the current settings in the fitting software for immediate use. The “Burn” button stores all the fitting parameters in the device EPROM memory. The “Disconnect” button terminates the connection between the hearing instruments and the computer.

The software also features a gain control slider, an output limiter slider, and an input-level slider. In the center of the fitting interface is an SPL-O-Gram that can display a wide range of data in the form of output dB SPL on the ordinate as a function of frequency on the abscissa. The check boxes on the right allow the experimenter to choose which data are displayed on the SPL-O-Gram at any one time. The lower center of the fitting interface displays a graphical equalizer used to manipulate the gain of the instrument as a function of frequency in 16 frequency bands denoted by their center frequencies in Hz or kHz. Buttons to the right provide global control over the equalizer parameters. In the lower left are the controls for the sound generator function. The sound generator is used in the fitting process to establish the RECD. For these measures, a white noise was filtered to include high-frequency emphasis that offsets (i.e., mimicked) the microphone noise floor, thereby providing a relatively constant signal-to-noise ratio while maintaining acceptable sound-level tolerance during the measurements. This selection is labeled “White” noise in the fitting software and is referred to as filtered white noise below. The sound generator also provides the therapeutic sound, which has a distinct spectral shape that is preconfigured, as discussed below. The slider allows the fitter to adjust the level of the sound generator output with buttons to the right for fine-tuning that level.

Reference measurements made on KEMAR and stored in the software can be selected and displayed in the fitting software. Similarly, after patient-specific measurements are completed and transferred to the fitting software, these data are available in the software for display on the SPL-O-Gram and are selected with corresponding check boxes. There are also options to display the patient-specific loudness discomfort levels (LDLs), pure-tone thresholds, and the standard minimum audible pressure curve (Dadson & King, 1952). Prior to fitting, LDL and threshold values are read into the fitting software from a database. A useful feature not available with the REM system is a display of the measured noise floor of the combined REM system and ambient room sound level. These data are made available in two ways. First, the noise floor measured using KEMAR is stored in the fitting software and available for view by selecting the appropriate check box. It is recorded by the REM system when there is no system sound presented and no live speech. Second, in the Session 1 fitting screen, there is an option to view the noise floor during each visit. In real time during fitting, the session noise floor is recorded and is shown on the REM system display along with the KEMAR referenced noise floor. If the current noise floor is substantially different from the reference noise floor, then the experimenter knows to investigate and eliminate the additional noise during that visit.

An overarching concern in fitting individuals with hyperacusis is their hypersensitivity to the overall loudness of all test signals that reach their ear canal. It is also easily shown that the spectrum of the probe microphone noise floor rises with frequency. To ensure that the lowest possible energy signal is used, while maintaining an adequate signal-to-noise ratio required for high confidence in the fitting measurements, a custom-shaped noise was created by filtering white noise as required. The resulting noise shape is both acceptable to study participants with LH and yields quality measurement data by increasing the probability that the noise, when presented at low levels, exceeded the measurement system noise floor.

Initiate Interactive Fitting Process

Prior to donning the device and initiating the fitting process, a data file is created with the ear and frequency-specific pure-tone thresholds and LDL values (measured following Sherlock & Formby, 2005) for the participant. Selection of the Open Session button in the fitting software loads this information into the fitting software. Next, selection of the Create Session 1 button creates a session file to be read by the REM system that contains the patient-specific information along with the parameters required for the first four REM system measurements. The REM software system “Restore Session” option is used to read the Session 1 information into the REM system. This data transfer procedure is repeated for Sessions 2, 3, and 4, which are detailed below. There are seven basic steps to the fitting process, as outlined in Table 1, that are simple in concept and mirror many of the steps involved in standard HA fitting, yet specific details associated with each step must be considered to accommodate hyperacusis and minimize the chance of overstimulation.

Table 1.

Fitting steps for each ear.

Measurement setup Measure and store equalization Measure REUR Measure REOR Set the SG level Program to unity gain Set the LS level

Note. REUR = real-ear unaided response; REOR = real-ear occluded response; SG = sound generator; LS = loudness suppression.

Measurement System Setup

After the devices are connected to the computer via the programming interface, the patient is seated in the center of a sound-attenuating chamber facing to the right of the chamber window. The on-ear assembly of the REM system is placed on each ear by looping the retention cord over the pinna and cinching it to secure the measurement apparatus below the ears with reference microphone positioned below the pinna and facing outward from the body. The probe microphone is positioned in the ear canal of the ear to be tested, while the other ear is occluded with the custom earpiece (without the embedded probe tube) and that treatment device is deactivated. Thus, the latter earpiece functions as a true earplug to avoid overstimulation of that ear when the test sounds are delivered in sound field from the REM system. The speaker and video monitor unit of the REM system is mounted on an articulating arm that extends from the wall of the sound chamber. During measurement, the unit is positioned approximately 18 in. from the patient with the speakers at a height equal to the opening of the ear canals. During measurement, the experimenter is seated on the control side of the sound booth window. The experimenter and patient can communicate via two-way talk-back system featuring microphones and speakers inside and outside of the chamber.

Stored Equalization

When making REMs, REM systems monitor the test signal sound pressure level (SPL) as a function of frequency using a reference microphone near to, but positioned outside of, the ear canal opening. This monitoring allows the experimenter to adjust the stimulus being presented, accounting for variations in that frequency response. The Audioscan Verifit 2 REM system defaults to the modified-pressure method with concurrent equalization, meaning equalization is completed prior to every stimulus presentation. Concurrent equalization considers the immediate acoustic conditions in the environment and patient-generated sound. However, concurrent equalization is problematic for open-ear responses because sound delivered by a hearing instrument in an unoccluded fitting has a tendency to leak out of the ear canal. The leaked sound will combine with the stimulus presented from the REM system, which will result in inaccurate equalization. To avoid this problem, the REM system has an optional modified pressure method with stored equalization, in which equalization is completed once, stored, and reused for each stimulus presentation. An added benefit of the stored equalization method is that repeated presentation of the equalizing stimulus prior to each measurement and recording can be avoided. In patients with hyperacusis, this reduces the number of times the patient is exposed to a potentially offending noise burst. The well-controlled measurement environment of a sound booth, along with careful monitoring of participant position and self-generated sound during measurement, helps to maintain constant conditions during a testing session. This consistency alleviates the need for updated measures via concurrent equalization. To use the stored equalization method, the REM system is placed into the “open” rather than the “occluding” on-ear mode so that the experimenter is prompted to perform the manual equalization. The Speech-ISTS stimulus option is chosen for simultaneous playback, and the recorded sound is stored for subsequent equalization (Byrne et al., 1994).

The REUR

The next measurement performed is the REUR. The REUR provides an output measurement as a function of frequency for a given input SPL. This measurement reflects the frequency-dependent filtering that the ear canal resonance characteristics impart on sound in the ear canal. The resulting transfer function ultimately serves as the target for the unity gain condition. The REUR is measured using a filtered white-noise stimulus presented at a level of 60 dB SPL. The spectral shape of the filtered white noise was designed such that the resulting ear canal sound pressure would approximate the shape of the probe microphone noise floor using knowledge of the average REUG. The resulting measured curve is stored, replacing the ISTS curve that had been stored in the equalization procedure above. Typical REURs show a resonant peak near 3000 Hz; however, the actual center frequency, level, shape, and slopes of the peak vary considerably from one ear to the next. When measuring the REUR, the probe tube is inserted into the ear canal to a depth corresponding to approximately 5 mm from the eardrum. To do so, the length of probe tube beyond the tragus is marked with the black ring that slides along the probe tube. The reference for positioning the black ring is based on that for the earmold with the embedded probe tube. The distance from the black ring positioned at the tragus to the end of the probe tube is measured, and that distance is transferred to the probe tube used in REUR. Placement is verified with otoscopy. In practice, the Audioscan ProbeGuide feature can be used to verify success of the placement procedure; however, that feature was not available when this protocol was initially developed.

All subsequent measurements are completed with the test ear occluded using the custom-occluding earpiece with an embedded probe tube. The non–test ear is occluded with the custom earpiece, but without the embedded probe tube. The latter minimizes the possibility of overstimulating the non–test ear when the test sounds are presented by the REM system. After verifying that both devices are muted (corresponding to a red background for buttons labeled “AMP”), the earmold is inserted into the test ear. We allocate 1–2 min for the stented earplug to reach body temperature, allowing the earplug to expand to conform snuggly to the ear canals.² Occlusion and probe tube patency then are confirmed by measuring the noise floor and comparing it to the previously recorded noise floor in the subsequent step.

REM System Noise Floor in Situ

Our use of low stimulus levels to avoid unnecessarily loud sounds is counter to the typical goal of high-level sound presentation to overcome occlusion due to an earpiece and the system noise floor (i.e., the combined noise of the REM system and test environment). The validity of measures made with low stimulus levels is dependent on the system noise floor. To measure the system noise floor, the occluding earpieces remain in the ears, the devices are muted, no stimulus is presented from the REM system, and an REM is captured. The resulting real-time noise floor measure is compared to the stored-reference noise floor measured with the KEMAR setup. In contrast, KEMAR uses half-inch microphones, which have significantly lower self-noise than the probe tube microphones found in REM systems. This comparison enables higher quality measurement data based on the manikin. The fitting software models the HA performance based on KEMAR measurements, which are then corrected to account for individual subject ear canal acoustic effects while monitoring the system noise floor.

Calibration Noise

As previously noted, the ear canal acoustic impedance of any individual subject can vary from the acoustic load presented by KEMAR to the HA output. To quantify this difference, an internally generated shaped white noise was developed that mimicked the probe mic noise floor measured on KEMAR. Labeled as “White” within the sound generator options of the fitting software, this option is selected as the source. The sound generator is enabled (unmuted) by pressing the red “SG” button, which then switches to green. The next measurement with the REM system records the level of this sound generator noise in situ as a calibration noise. The sound generator is then muted. The fitting software utilizes the difference between the in situ measured calibration noise and its corresponding KEMAR data as the basis for the corrections needed to improve the predicted performance.

REOR

The next step is to measure the REOR to establish the frequency-dependent attenuation achieved with the earpieces properly inserted into the ears. The REOR measurement yields a response curve representing output in dB SPL as a function of frequency. The REOR then can be compared to the REUR to establish the amount of attenuation or insertion loss provided by the device in situ. Prior to the REOR measurements, the patient is instructed that a brief, loud sound will be presented. The experimenter confirms that both earmolds are in place, properly fit, and that the devices are muted before the measurement stimulus is presented (to avoid overstimulation). Typically, a single relatively high stimulus presentation level would be used to capture the attenuation of the earpieces. For some individuals, however, higher presentation levels may not be tolerable given their abnormal sound sensitivity, especially if the earpiece provides less than desirable attenuation (which is unknown at this point in the fitting procedure). Thus, a step-up procedure is used in which an initial measurement is conducted using a white noise stimulus presented at a level of 60 dB SPL. For some individuals, this lower presentation level can fully capture the attenuation of the earpieces, and no other measurement is required. For other individuals, this level might be insufficient to capture attenuation across all frequencies. Such testing at this lower level provides an accurate indication that the device is providing the anticipated attenuation without putting the participant at risk for overstimulation. Figure 7 illustrates an example showing that the measured REOR is less than the REUR at all frequencies tested. The presence of good attenuation in the low-frequency range indicates that there is a good seal of the ear canal and that no slit leaks are present. The fact that the REOR lies in very close proximity to the probe mic noise floor (shown in dark blue) indicates that, although we have a sealed ear and attenuation attributable to the earmold, we do not know the full degree of in situ REOG for the system. Thus, after this initial REOR measurement with white noise at 60 dB SPL, which provides confidence that the subject will not be exposed to stressful sound levels, a second REOR is recorded for pink noise presented over a shorter duration at 80 dB SPL. The use of a higher level stimulus allows for an improved estimate of the actual REOG. The measurement data are saved by the REM system and read into the fitting software for display and calculation purposes. Each REOR measure is compared to corresponding KEMAR measures to check for any substantial discrepancies.

Model Flat REIG

The difference between the REOR and REUR measurements, corrected for the differing stimuli spectra used in each measurement, provides the REOG. This gain reflects the attenuation achieved with the earpiece and quantifies the transfer function of the direct sound path of the underlying acoustic model. Likewise, the results of the calibration noise measurement allow for a more accurate prediction of the amplified path. As a first approximation to achieving a flat REIG measurement, REAG was calculated based on the acoustic model described above (shown in Figure 5) as the power sum of the amplified and direct paths followed by correcting for the individual REUG. Based on this calculation, changes to the HA gain and equalization parameters are made automatically by the fitting software to target a unity gain flat REIG. The resulting prediction of HA performance is displayed on the SPL-O-Gram. If the REIG is not flat, then there is an observable mismatch between the REAG and REUR curves. The experimenter then can and should make further manual adjustments.

REAR and Flat REIG

With the optimized parameters uploaded to the HA, the hearing device in the test ear is unmuted (the AMP button is selected, changing the button color from gray to green), allowing the stimuli captured by the microphones to be amplified. The same-shaped white noise used for measuring REUR is presented at 60 dB SPL for the REAR measurement. The resulting REAR curve is displayed in the REM system window, along with the REUR from the prior measurement. If the model predictions are perfectly matched, then the REAR curve should overlay the REUR curve and the corresponding REIG should be flat. If not, then the sliders in the equalizer portion of the fitting software are used to match the REAR to the REUR. This match of the aided and unaided responses represents unity gain. When unity gain is achieved, the hearing device is functioning in a transparent mode, effectively offsetting the attenuation and ear canal resonance alterations (i.e., insertion loss) produced by the snugly fitted stented earpiece. After this match is complete, the amplification was muted once again.

Setting of the Therapeutic Noise Level

To set the therapeutic noise level from the sound generators, the radio button labeled “Therapy” is chosen in the fitting software. This applies a predetermined filter to the native sound generator noise to achieve the desired therapeutic noise. The latter closely corresponds to the therapeutic noise used in prior investigations (e.g., Formby et al., 2015). Prior to activating the sound generator, the user is informed that they will hear a “seashell”-like sound that will gradually be increased in level from inaudible until they judge the loudness to be “soft but comfortable.” The sound generator output then will be increased further until they judge the sound to be “comfortable” in loudness. These loudness judgments are based on contour test Categories 3 and 4 (Cox et al., 1997), which were familiar categorical judgments for the study participants in our field trial (Formby, Cherri, et al., 2024).

The experimenter begins the procedure by setting the corresponding fitting software slider adjustment to the lowest level available. Then, the therapy noise is enabled using the SG button in the fitting software. The sound generator level is increased from the minimum level in 2-dB steps until it is judged to be a Category 3 “soft, but comfortable,” and then the level is increased in 1-dB steps until it is judged to be a Category 4 “comfortable.” This process is repeated 3 times, and the median value among all Category 3 values is set as the sound generator level. The final sound generator level is then presented to the participant, who is asked to confirm that this level is between a Category 3 and Category 4. Once confirmed, the sound generator output is recorded in the ear canal, and the measured frequency response is displayed on the screen. The therapeutic sound is then muted for the remainder of the fitting measures and reactivated when the fitting is finalized. The probe microphone noise floor typically is only slightly below the level of the sound generator curve, which makes the sound generator measurement in the prior step vulnerable to the system noise floor. Thus, it is important to confirm visually the validity of the sound generator output curve in comparison to the noise floor curve. As previously described, the in situ calibration noise measurement is used to improve the accuracy of the RECD component within the multipath acoustic model. The KEMAR measured spectrum of the therapeutic noise, displayed within the fitting software, also inherits improved predictive accuracy from the process. Thus, the experimenter can use this augmented knowledge to gain confidence about the therapy noise being provided in cases where the direct REM measurement is considered unreliable.

LS: Setting Output Limiting

The setting of the output-limiting LS activation level effectively restricts the device output to sound levels below patient-specific Category 6 threshold judgments for “loud, but OK.” The desired LS activation-level setting, corresponding to “loud, but OK” judgments, is approached in a gradual manner with several safeguards in place to avoid overstimulation. First, the device output limit control is adjusted to its lowest possible value (35 dB SPL). Next, the device amplification is enabled, restoring device amplification to unity gain. This results in the least possible output from the device when activated. The input-level control in the fitting software then is set to the maximum value (90 dB SPL) to provide an updated estimate of the maximum device output, which when compared visually to the subject's LDL data ensures a first assessment of the appropriateness of the device capabilities. Once verified, continuous speech is then delivered at an output level of 50 dB SPL using the prerecorded “carrot passage” spoken by a female talker available in the REM system. With the device output-limiting set at the lowest possible level and the carrot passage presented in a loop, the REM system presentation level is increased gradually in 5-dB steps up to a presentation level of 80 dB SPL (or in rare cases, as close to 80 dB as can be achieved before the user judges the running speech to be “loud, but OK”). During these steps, the LS activation-level setting remains at 35 dB, thereby maximally limiting the device output level. After the highest acceptable presentation level of the carrot passage is achieved, the LS activation-level setting (i.e., the device output-limiting control setting) then is increased gradually from 35 dB SPL in 1- or 2-dB steps until the participant initially judges the level of the running speech as “loud, but OK.” The output-limiting LS setting then remains set at this level, while the input level to the device is decreased back to its initial conversational level.

This concludes all the steps needed to achieve unity gain, the output-limiting LS setting, and the setting of the therapeutic noise level for the left ear. Each of these steps is then repeated for the right ear, with nominal adjustments made afterward if necessary to provide balanced perception between the ears. Prior to saving and disconnecting each device, the amplification and therapeutic sound generator are enabled, and a second program is created, which has both the amplification and sound generator disabled. This mute program allows some peace of mind for the user, who in particularly challenging environments can switch to the full benefit of the earplug if desired. This protective safeguard feature increases their willingness to use the devices in environments that they have been avoiding. The steps for setting LS are repeated at approximately monthly intervals to allow for increases in the LS setting as improvements in sound tolerance are achieved over time.

Future Directions, Limitations, and Alternatives

The transitional intervention for hyperacusis described here consists of a unique combination of several concepts including counseling, sound limiting, enriched exposure to sound, and controlled transitioning of a patient from one mode of sound reception to another in a way that minimizes the potential for discomfort. A major advancement, when successfully implemented (see Formby, Secor, et al., 2024), is the transitioning of patients with LH from their counterproductive sound avoidance behaviors (i.e., sound attenuation and limited exposure to healthy low-level sounds) into beneficial treatment with therapeutic sound. This treatment-driven transition is accomplished with an active hearing device that combines the effects of earplug attenuation with unity gain and output-limiting LS. A second major advancement is the precision with which the devices can be fit to achieve unity gain and the associated fitting process, which is designed to do no harm. A third major advancement is the adaptive nature of the fitting, in terms of dynamic occlusion (i.e., open dome vs. custom occluding earpiece), which optimizes the use of active sound therapy in controlled and uncontrolled sound environments, and the treatment-induced changes in LS, which progressively facilitate and maximize exposure to the higher healthy environmental sound levels that offer passive sound therapy.

A key to the success of this transitional intervention will be successful commercialization of the technology and the fitting protocol. It is notable that various analog iterations of loudness-suppressing devices for LH were previously described by Preves et al. (1995), Nunley (1996), Sammeth et al. (2000), and Valente et al. (2000). The devices reported by Preves et al. and Sammeth et al. were manufactured by Argosy, which, by 2000, had discontinued production of their device (Sammeth et al., 2000). MicroTech commercialized a similar “electronic attenuator” product that they labeled the “Refuge” (Valente et al., 2000), which also was subsequently discontinued. An analog Class D amplifier instrument, initially designed by Nunley, labeled the “Star 2000,” was later redesigned by Fenwick as a digital product, the “Star 2001” (Vernon, 2002; Vernon et al., 2002), for protective management of the patient with LH. Despite promising anecdotal reports of success with these various devices (most notably improved communication and increased sound protection) in some situations for some hyperacusis patients (described in case examples; see Westcott, 2006, and the reports above), none of these products were evaluated formally in rigorous prospective studies. Moreover, none of these products ultimately achieved widespread use, and to our knowledge, all have long since been discontinued.

Although we do not know the various reasons for the failure to commercialize and advance the development of these products further, we suspect one factor may have been uncertainty at that time about the prevalence of hyperacusis and, thus, the expectation of low sales of these products. Another factor, which is apparent from this report, was the myriad technical challenges of achieving and fitting these devices. Moreover, very few clinicians and clinical centers were well versed in the management and treatment of hyperacusis at the turn of the century, and this paucity of competence almost certainly was a limiting factor in the promotion and adoption of these products. Contrary to the narrative in some of the reports above, these loudness-suppressing devices afforded no treatment per se for hyperacusis. Indeed, these products were designed as a prosthetic for enhanced protection of the patient, with the added advantage of affording a larger dynamic range than otherwise would have been achieved with passive HPDs. Perhaps this confusion, vis-à-vis the false expectation that these devices afforded a primary treatment benefit, may also have contributed to failure to adopt the technology.

Perhaps another limitation is that the devices and protocol described here were limited in terms of user adjustment. In theory, the gain could be adjusted to values less than unity gain to achieve some desired amount of attenuation or earplug effect. The fitting protocol described here does not provide access to a user volume control to reduce gain relative to unity, or a volume control for setting the output levels from the sound generators. This decision to eliminate these optional user features from the intervention device was solely based on research considerations, enabling Formby, Cherri, et al. (2024) better control of sound exposures and device use conditions during the associated field trial of the transitional intervention. In a commercial product, such features could be included.

The device fitting and adjustment protocol described here was developed for research purposes and includes several steps that can be simplified. This includes leveraging a more streamlined closed-loop fitting procedure through the linkage of the REM system and fitting software. Such streamlining could eliminate much of the manual control required to execute the steps described above. This automation is possible now as a result of standard communication protocols; however; it was not possible at the time of the study inception. Likewise, additional features that likely will become available to REM systems, including noise floor measures, will simplify the steps needed to ensure accuracy while minimizing sound exposure to the patient. The initial fitting process, following the procedures outlined above, took about 1 hr to complete. With the changes noted above, a goal would be to reduce that time to approximately 20–30 min. The device, as described here and designed for the initial field trial, requires all adjustments to be made in the laboratory or clinic. The number of visits required in the current protocol is very resource heavy from the perspective of the health care professional and somewhat inconvenient from the patient perspective. Thus, future developments will seek to determine the optimal number of visits, which will likely include some combination of in-field evaluation and device control via software applications accessible to the end user remotely. Benefits also may accrue from one or more forms of teleaudiology and data-sharing applications.

Conclusions

The device and fitting protocol outlined here were designed as part of an initial field trial of a transitional treatment for debilitating hyperacusis (see Formby, Cherri, et al., 2024). The key features (i.e., therapeutic sound generators, LS, and different sets of earpieces that provide comfort without sacrificing benefit) for the intervention are integrated in a single protective treatment device. With additional research and more widespread clinical use, the device and the fitting and adjustment protocols for the output-limiting sound protection and the therapeutic sound components likely can and will be streamlined. Ideally, these and similar protocols, including our companion counseling protocol (Cherri et al., 2024), will become part of best practices guidelines. Our protocols therefore have the potential to be incorporated in modern hearing devices and made available for treatment of people with hyperacusis and other reduced sound tolerance conditions. Such a process of bringing new interventions and methodologies to the clinical forefront typically spans decades in the field of audiology. Thus, sharing these procedures early in the evolution of our treatment protocol hopefully will stimulate interest and spur replication and expansion of the core research that has led to the development of the novel procedures described in this report.

The acoustic model used within the fitting software, in conjunction with high-quality data measured on a KEMAR manikin, resulted in a flexible framework that accounts for venting effects. When updated with individual subject relative differences, the model proved very effective in our research and could offer the basis for more general use.

Appendix A

Earmold Impressions

In our field trial (Formby, Cherri, et al., 2024), two ear impressions were created for each ear: one without a probe tube in the ear and one with the probe tube in the ear. The goal of the second impression was to allow the manufacturer to produce an earpiece with a probe tube permanently embedded at the correct insertion depth for in-ear measures to ensure a good acoustic seal and to eliminate the possibility of acoustic (slit) leaks that likely would occur if the probe tube were placed between the ear canal wall and the earmold itself. The first impression was obtained using standard clinical procedures. During the second impression procedure, a probe tube was placed within 5 mm of the tympanic membrane, as confirmed with otoscopy. The outer portion of the probe tube was secured to the pinna with medical tape while the inner portion of the probe tube was secured in the desired position against the canal wall when the otoblock was carefully inserted over it. Otoscopy was repeated to confirm that otoblock placement had not shifted the position of the probe tube. The impression was then obtained with the probe tube remaining in place in the ear canal. After the impression was removed from the ear canal, it was marked precisely by the audiologist for desired length of the earmold canal and depth of the probe tube. Two pairs of custom earpieces were delivered from the manufacturer: a pair without probe tubes for standard use by the participant and a second pair with embedded probe tubes for laboratory fitting purposes. During the initial fitting process with each participant, unoccluded probe microphone measurements required the use of a separate probe tube. The goal was to match the placement of that probe tube as closely as possible to the position of the embedded probe tube. Commercially available probe tubes have a moveable marker that can slide along the length of the tube to mark canal depth. This feature allows an audiologist to insert the probe tube to the same depth each time it is inserted. Accordingly, prior to completing the unoccluded probe microphone measures in the initial fitting procedure, the position of the depth marker on the embedded probe tube was aligned with that on the separate probe tube to ensure that the unoccluded and occluded measures were all obtained at the same probe mic depth.

Appendix B

Summary of Device Fitting Protocol

Device fitting protocol
Action	Description
Obtain 2 sets of well-fitting earpieces per ear, one with embedded probe microphone tube	The earpieces must be able to function as high-quality earplugs when coupled to the device. The use of a duplicate earpiece with embedded probe microphone (the fitting earpiece) allows for real-ear measures to be completed during the fitting visit without compromising the attenuation of the earpiece (via introduction of a slit leak). This process is detailed in Appendix A.
Occlude the ear that is not being fit	This provides protection to the ear while the device for the opposite ear is programmed.
Insert REM probe microphone into the ear canal of the ear being fit and measure REUR	Accurate placement of the probe microphone and verify that it is positioned 5 mm from the tympanic membrane prior to measuring the unoccluded response.
Insert the device, coupled to the fitting earmold	Ensure that the embedded probe microphone is not bent or pinched during insertion. Allow at least 1 min for the stented earpiece to provide full expansion.
Measure the REOR	Measure the response of the occluded ear canal. The difference between REUR and REOR is the passive attenuation of the earpiece. This measure allows for the calculation of insertion loss and also confirms the effectiveness of the earpiece before proceeding.
“Flatten” REIG to estimate the settings needed to achieve unity gain	Model calculations are used to rapidly approximate flat REIG. Fine tuning will be completed in a later step.
Set the SG level	Enable the SG and increase the SG setting in 2-dB steps until a comfortable level is reached.
Program to achieve unity gain Match the REAR to the REUR	The software calculates the difference in the acoustic response due to the insertion of the occluding earpiece (insertion loss). It applies the calculated differences across the frequency bands to correct for the insertion loss and approximate unity gain (REAR = REUR). Make additional manual adjustments as needed.
Set the LS	To determine the LS setting, begin with the device muted, then play running speech in the sound field at a low level. Unmute the device, then gradually increase the presentation level to 80 dB SPL, or as close to that level as tolerated. Once the stimulus is at or near 80 dB SPL, begin to increase the LS slider in 1- or 2-dB steps to allow more sound to pass through the device. Continue until the participant indicates that the running is perceived as “loud, but OK.”
Create a 2nd program in which both amplification and SG are muted	The mute program allows the device to function solely as a highly effective earplug, if needed.
Repeat the process in the other ear

Note. REM = real-ear measurement; REUR = real-ear unaided response; REOR = real-ear occluded response; REIG = real-ear insertion gain; SG = sound generator; REAR = real-ear aided response; LS = loudness suppression.

Funding Statement

The preparation of this report was supported by an award from the National Institutes of Health (National Institute on Deafness and Other Communication Disorders R21DC015054) to C. Formby and D. A. Eddins.