Ampclusion,1 or the “hollowness” and “unnaturalness” of a hearing aid wearer’s own voice during vocalization, is a frequent complaint among hearing aid users. There are many factors contributing to this complaint; the two primary ones include a shell origin and an amplifier origin.2,3 The shell origin has its source from the physical occlusion of the ear-canal with hearing aid use. The amplifier origin has its source from sub-optimal (too much, insufficient, or simply different from previous experience) frequency-gain characteristics of the hearing aids, saturation distortions and processing artifacts (eg, group delays in digital hearing aids) of the hearing aids.

A 5-step protocol4 was proposed to help clinicians involved in hearing aid selection and fitting manage this common complaint. Although many ampclusion complaints involve a mixture of both shell and amplifier origins, it is highly likely that the ampclusion complaint in someone with a high frequency hearing loss (HFHL) originates from shell occlusion alone, if one assumes that the ampclusion complaint is a result of excessive low frequency SPL in the ear-canal. This observation has a significant impact on the management of this hearing loss configuration. The following article provides an explanation and the clinical implications of such a speculation.

How is Objective “Hollowness” Measured?
Scientists and clinicians have devised both objective and subjective measures of the “hollowness” complaint. Researchers typically measure the Occlusion Effect (OE) in order to objectively quantify the magnitude of the wearers’ hollowness complaint. The wearer is asked to vocalize /i/ with and without the hearing aids in-situ; the real-ear unaided response during vocalization (REURvoc), the real-ear aided response during vocalization (REARvoc), and the real-ear occluded response during vocalization (REORvoc) are typically measured. The difference in real-ear output below 1000 Hz between the REURvoc and the REORvoc is defined as the OE.5 Similarly, the difference between the REURvoc and REARvoc is the Ampclusion Effect (AE). In principle, the AE includes the OE and the contribution of amplification from the hearing aid. Figure 1 shows a sample of the real-ear output of the three responses. In this case, the OE at 400 Hz is calculated to be around 20 dB, while the AE is calculated to be 25 dB.

figure
FIGURE 1. Typical real-ear responses during vocalization of /i/. The REURvoc is typified by the ear canal resonance. The REORvoc is typified by the increase in SPL below 1000 Hz and the abolition of the ear-canal resonance. The REARvoc is typified by an increase in real-ear sound pressure level and the restoration of the ear-canal resonance.

One way to interpret the findings in Figure 1 is that the magnitude of the OE (about 20 dB) represents the contribution of the shell occlusion alone to the whole percept of hollowness or unnaturalness. On the other hand, the magnitude of the AE (about 25 dB) would represent the combined contribution of shell occlusion (i.e., OE) and amplification provided by the hearing aid of the wearer’s own voice during vocalization. If the magnitude of the AE is the same as the OE, one would conclude that the ampclusion complaint is primarily determined by the shell occlusion. If the gain settings of the hearing aid contributed to the real-ear output, one would expect the magnitude of the AE to be higher than the OE.

REARvoc as a Function of Low Frequency Hearing Loss
In order to determine the extent of the low frequency hearing loss that would result in a REARvoc that is higher than the REORvoc, we measured the REARvoc and REORvoc (and REURvoc) with the Senso Diva SD-9 BTE hearing aid set for different degrees of hearing loss below 500 Hz (from 20 dB HL to 70 dB HL in 10 dB steps) but with the same degree of hearing loss of 70 dB HL at and beyond 1000 Hz (see inset in Figure 2).

figure
FIGURE 2. The Ampclusion Effect (AE) for different degrees of hearing loss at 500 Hz. Data for both the closed (blue line) and open vent (pink line) conditions are shown. The magnitude of the OE (REORvoc – REURvoc) is also indicated by the asterisks. The inset audiogram shows the different hearing loss configurations for setting the hearing aid gain.

A lucite skeleton earmold was used for coupling. The earmold had a Select-A-Vent system and it was used in the closed condition (ie, no vent) and the open condition (ie, 2 mm vent). After the measurement, the OE at 400 Hz for each vent condition (closed and 2 mm) was calculated by subtracting the REURvoc from the REORvoc at 400 Hz. The AE at 400 Hz for each configuration of hearing loss was calculated by subtracting the REURvoc from the REARvoc at 400 Hz for both vent conditions as well.

Figure 2 shows the magnitude of the AE as a function of the degree of hearing loss at 500 Hz (from 20-70 dB HL) for both the closed and open vent conditions. The magnitude of the OE (REORvoc – REURvoc) was also indicated in asterisk for comparison. In the closed-mold condition, the OE was measured to be 17 dB. The AE increased gradually to about 19 dB when the hearing loss was 40 dB HL. It increased more rapidly above a hearing loss of 60 dB HL to around 27 dB at a hearing loss of 70 dB HL. The vented earmold, as expected, yielded a lower OE of 10 dB. The magnitude of the AE stayed around 10 dB until the hearing loss at 500 Hz exceeded 30 dB HL. It increased to around 22 dB at a hearing loss of 70 dB HL.

These observations showed that, for the same degree of high frequency hearing loss, the REARvoc or AE is similar to the REORvoc or OE until the hearing loss at 500 Hz exceeds 30 to 40 dB HL. In other words, unless the hearing loss at 500 Hz of a person with a HFHL exceeds 30-40 dB HL, his/her complaint about hollowness or own voice being unnatural is most likely due to shell occlusion (ie, shell origin).

Why is Shell Occlusion the Culprit?
During vocalization, the sound pressure level that is generated by the bone-conduction mechanism escapes through the ear-canal. When the ear is occluded (as in hearing aid wear), the trapped SPL within the occluded ear-canal gives rise to the perception of the OE. When the hearing aid is turned on, the SPL trapped within the ear-canal (ie, OE) will be added (assuming that they are in phase and no cancellation occurs) to the amplified sounds (of the wearer’s own voice) from the hearing aid.

For simplicity, let’s use the real ear output at 400 Hz in Figure 1 as an example. Real-ear output from the bone conduction mechanism can be estimated from the REORvoc which is 85 dB SPL in this case. Real-ear output from the hearing aid is dependent on the level of the input signal and the gain appropriate for that input level. In the low frequencies, one may assume that the SPL measured in the ear-canal is similar to that measured at the hearing aid microphone. Figure 1 shows that the REURvoc at 400 Hz is about 65 dB SPL. If the gain at this frequency and input level is 10 dB, then the output from the hearing aid will be 75 dB SPL; if the gain is 20 dB, it will be 85 dB SPL and so on.

One must remember that the intensity of sounds is expressed in dB, which is a logarithmic representation of the sound pressure level of an acoustic event in reference to a standard (20 µPa). When one adds two sound sources together using dB notation, one needs to convert the sounds from dB SPL into the appropriate sound pressure (Pa) before adding them together. And once they are added, one has to convert them back into dB SPL. For example, a sound that is 100 dB SPL (re: 20 µPa) would have a sound pressure of 2 Pa. Thus, when a sound of the same magnitude is added, the combined sound will have a sound pressure of 4 Pa. When this is converted into dB, it would correspond to a dB level of 106 dB SPL. The dB notation limits the increase in overall sound pressure level to 6 dB over the sound pressure level of the louder sound when two sound sources are added.

Figure 3 shows the resulting overall dB SPL when two correlated sound sources are added together. For example, if two sound sources that are identical in sound pressure level (eg, 70 dB SPL, identified by “0” on the x-axis, meaning that the two sound sources are identical in sound pressure level) are added, the overall SPL will be 76 dB SPL. Thus, an increase of 6 dB is shown on the y-axis. If, on the other hand, one sound source is 70 dB SPL and the other is 60 dB SPL (ie, a 10 dB difference in sound pressure level), the total sound pressure level will be increased by 2.4 dB over the louder sound source. In this case, the resulting sound level will be 72.4 dB SPL. If one sound source is 70 dB SPL and the other is 50 dB SPL, the overall sound pressure level will only increase by 0.8 dB over the more intense sound to 70.8 dB SPL.

figure
FIGURE 3. Calculations to show the increase in overall sound pressure level (SPL) when two sound pressures are added together. The x-axis represents the difference in SPL between the two sound sources. The y-axis is the number of decibels that will need to be added to the more intense sound pressure level to arrive at the overall SPL. This estimation is appropriate for all input levels.

Consequently, if the REORvoc is 85 dB SPL, the output of the hearing aid must be at least 70-75 dB SPL before the output from the hearing aid would add 1.5-2.5 dB to the REARvoc to make it larger than the REORvoc. With an input sound pressure level of 65 dB SPL at 400 Hz, the hearing aid must provide at least 10 to 15 dB gain to make the REARvoc 1-2 dB higher than the REORvoc. More gain will be needed if a larger difference between the REARvoc and REORvoc is expected. This condition will only be realized until the hearing loss in the low frequency is substantial.

For individuals with a HFHL where the hearing loss in the low frequency is minimal, the REARvoc measured in the low frequency will be similar to the REORvoc. Since the REORvoc is a result of shell occlusion alone, the origin of the ampclusion complaint for people with a high frequency hearing loss is mainly shell origin.

Implications of Shell Origin
The speculation that ampclusion complaint in people with a high frequency hearing loss is primarily shell origin has important implications in the selection and fine-tuning of hearing aids. From the selection standpoint, this knowledge should direct one to choosing hearing aids that minimize shell occlusion. Kuk & Ludvigsen2 reported that the two common approaches to minimizing shell occlusion are: 1) to increase the leakage (by increasing vent diameter and/or shortening vent length) of the earmold or hearing aid shell and/or 2) to increase the insertion depth of the hearing aid shell/earmold.

Minimizing OE with Shell Origin
Kuk, Keenan & Lau (“Vent Configurations on Subjective and Objective Occlusion Measurements”; submitted paper, 2005) evaluated the changes in OE as the vent diameter of an ear insert that terminates at the second bend of the ear canal was modified. The ear inserts were made using laser-fit shell-making technology (CAMISHA)6 in order to ensure a precise seal of the shell in the ear-canal. Subjects were asked to monitor and vocalize /i/ for 5 seconds. The recorded real-ear unaided and occluded responses for the different vent conditions recorded during the middle 3 s of vocalization were frequency-averaged. The data were analyzed using regression analysis.

Figure 4 shows the modeled occlusion effect as the vent diameter was altered. For example, for an insert that has the typical CIC length of 16 mm, the average wearer will show a 5 dB reduction in OE as the vent diameter is increased from 0.5 mm to 1 mm. Further increase in vent diameter leads to less incremental change in OE. For example, less than 2 dB decrease in OE is noted as the vent diameter increases from 2 mm to 2.5 mm. Shortening of the vent length would also lower the OE.7

figure
FIGURE 4. Calculated Occlusion Effect (OE) as a function of vent diameter (in mm) for a 16-mm long ear insert.

In recent years, there is a re-kindled interest in the use of tube-fitting (or open-fitting) to minimize or eliminate OE. A miniature behind-the-ear hearing aid is coupled to either an ear-set or a standard size #13 tubing for the delivery of amplified sounds to the wearer’s ears. Because the ear canal is totally unoccluded, advantages of open-fitting include no or minimal occlusion effect, preservation of natural ear canal resonance, natural sound quality and improved localization ability. On the other hand, open-fitting limits the usable gain on a hearing aid and may compromise the effectiveness of the signal processing algorithms employed in that hearing aid.

Fortunately, with careful design considerations, an open-fit hearing aid can eliminate occlusion with only minimal compromises on the effectiveness of the signal processing and audibility provided by the hearing aid. For example, Kuk et al.8 reported that the Diva élan hearing aid (an open-fit hearing aid with a fully adaptive directional microphone and noise reduction algorithm) was able to eliminate all objective OE while providing its wearers a signal-to-noise ratio (SNR) advantage of almost 3 dB over the omnidirectional aided condition with no noise reduction. This type of fitting may be especially beneficial to people with a precipitous high frequency hearing loss.

Another seemingly opposite approach to minimizing the OE is to increase the insertion depth of the earmold/hearing aid shell to beyond the second bend of the ear canal (bony portion) so that the cartilaginous portion of the ear-canal will not vibrate (ie, the vibration produces the OE). Revit5 showed that the OE was reduced to only 2-3 dB when an ear insert was inserted close to the eardrum. Pirzanski9 showed that the depth of insertion of the shell/insert must be at least 4 mm beyond the second bend of the ear-canal for the wearer to experience a substantial reduction in OE. Physical comfort may be a concern to some wearers using this approach to minimize OE.

Fine-tuning Hearing Aids
Fine-tuning efforts for ampclusion complaints in high frequency hearing loss must be directed to improving the shell occlusion, and not to modifying the electroacoustic settings of the hearing aids. However, many clinicians still adjust the low frequency gain on the hearing aids in an attempt to minimize AE in wearers with a HFHL. Others activate the “occlusion manager” that is available on some digital hearing aids to manage the hollowness complaint. Unfortunately, the current “occlusion managers” are dedicated only to AE that has an amplifier contribution. This is because most “occlusion managers” only allow adjustment of low frequency gain on the hearing aid.

For example, Figure 5a (left) and 5b (right) show the simulated real-ear output of the Diva hearing aid (with a closed earmold) to sounds presented at input levels of 40 dB, 65 dB, and 90 dB SPL for a high frequency hearing loss of 20 dB HL at 500 Hz and 70 dB HL from 1000 to 4000 Hz. Figure 5a shows the output at the default setting, while Figure 5b shows the influence of the Occlusion Manager in addition to maximum gain reduction (for all input levels) at and below 500 Hz. The latter (Figure 5b) approximates the extreme situation where all the low frequency gain on the hearing aid is removed. One can see a significant difference in frequency-output characteristics between the two conditions, especially with the specificity of gain adjustment provided by the Diva Occlusion Manager seen in steepness of response below 1000 Hz in Figure 5b. Perceptually, the default settings on the Diva aid would sound different than the adjusted settings to environmental sounds.

figure
FIGURE 5A-B. Frequency output curves to input levels of 40 dB (bottom), 65 dB (middle), and 90 dB SPL (top) for a high-frequency hearing loss with thresholds of 20 dB HL at 500 Hz and 70 dB HL at and above 1000 Hz. Figure 5a (left) shows the default frequency output characteristics. Figure 5b (right) shows the frequency output when the occlusion manager was activated and all the low frequency gain was set to minimum.

On the other hand, Figure 6 is the corresponding REARvoc for the default setting, the adjusted setting with only the occlusion manager activated, and the adjusted setting with the occlusion manager activated and low frequency gain minimized. The REORvoc was also included for comparison. As one can see, the REARvoc for all three gain settings are almost identical. More importantly, the REARvoc are identical to the REORvoc, suggesting that all the low frequency adjustment (in occlusion manager activation and insertion gain adjustment) was not effective in reducing OE that has a shell origin.

figure
FIGURE 6. Real-ear aided response during vocalization (REARvoc) at the default setting (red), with activation of the occlusion manager alone (green), and with activation of the occlusion manager plus low-frequency gain reduction (turquoise). The REORvoc (dark blue) was also included for comparison. All four real-ear output curves were identical below 800 Hz, confirming the ineffectiveness of low-frequency gain adjustment for ampclusion complaint that has a shell origin.

Worse still, the audibility of speech sounds and the sound quality of everyday sounds may also be compromised from this action.10 This example stresses the importance of a correct diagnosis (ie, shell origin of ampclusion in HFHL) and the right solution (ie, shell modification and not gain adjustment) in fine-tuning ampclusion complaints.

Variations to Speculation
The previous discussion on the shell origin of ampclusion complaints in people with a HFHL is based on the assumption that the ampclusion complaint is simply the result of extra or excessive low frequency SPL in the ear-canal over the unaided state. In real-life, there may be other variables that can alter the strength of this speculation.

Experience and expectations. The term “excessive” low frequency energy may be subjective. The same amount of low frequency energy in the ear canal may be perceived differently by two wearers with identical hearing losses but with different levels of experience in using a hearing aid. For example, an experienced hearing aid wearer would less likely complain about ampclusion than a new hearing aid wearer. Thus, counseling and setting the appropriate expectations for new hearing aid wearers are important first steps.

Hearing aid gain in the low frequencies. A high frequency hearing loss includes people with a precipitous hearing loss (ie, normal hearing in the low-to-mid frequencies), as well as those with a mild-to-moderate degree of hearing loss in the low-to-mid frequencies. Obviously, the more hearing loss there is in the low frequencies, the more gain will be prescribed, and the more likely that the ampclusion complaint will have an amplifier contribution. Figure 2 suggests that, for a hearing loss with 40 dB HL (or greater) at 500 Hz, an ampclusion complaint would likely be affected by low-frequency gain settings on the hearing aids. Conversely, a hearing loss less than 30 dB HL would most likely have a shell origin to the ampclusion complaint.

Similarly, people with a HFHL who wear hearing aids fit with prescriptive formulae that recommend more gain in the low frequencies than targets that specify less gain would more likely have ampclusion complaints that have an amplifier contribution (in addition to having shell occlusion). This is because the higher gain prescription would result in a higher REARvoc. For hearing aids fitted using such a prescriptive target, it is likely that the lower limit of hearing loss at 500 Hz may be 20-30 dB HL for the REARvoc to have significant amplifier contribution.

Signal processing. In line with the assumption of excessive SPL as the source of ampclusion complaint, one would speculate that signal processing approaches that provide more gain to high input levels (such as the wearer’s own voice) would have a higher REARvoc than those that provide less gain. If one recalls that a wide dynamic range compression (WDRC) hearing aid provides more gain for soft sounds and less gain for loud sounds than a linear hearing aid (which has the same gain for all conversational input), one would imagine that a linear hearing aid will result in a higher REARvoc than a WDRC hearing aid. In other words, there is a higher likelihood that the ampclusion complaint for wearers of a linear hearing aid to have amplifier contribution as well.

Indeed, Kuk et al.11 showed that a “noise reduction” hearing aid yielded less hollowness perception than a linear hearing aid during vocalization because of the automatic gain reduction from the noise reduction algorithm of the hearing aid. Figure 7 compares the REARvoc (or AE) between the Diva hearing aid set at its default (WDRC) and with it set to linear processing (with the same gain for a conversational input). One can see that the REARvoc in the linear case was higher than the default WDRC processing because more gain for high input levels is provided in the linear case. In addition, the AE with the linear processing was about 3-5 dB higher than that of the OE for all degrees of hearing loss at 500 Hz.

figure
FIGURE 7. Ampclusion effect (at 400 Hz) between the Diva hearing aid set to its default WDRC processing (blue line) and linear processing (red line), while keeping the same gain for a conversational input. The magnitude of the REORvoc is also indicated.

This suggests that the AE reported by the wearer of a linear hearing aid could have a larger amplifier contribution than that of a nonlinear hearing aid. Although the majority of today’s hearing aids are nonlinear, it is important to acknowledge this possibility during fine-tuning of hearing aids for hollowness complaints.

Other artifacts. It was indicated previously that some wearers report artifacts of the hearing aids during their vocalization as “hollowness” or “unnatural voice.”2 Factors such as the group delay of the hearing aid, saturation distortion occurring at the input stage and/or the output stage of the hearing aid during vocalization, and too much compression (yielding a muffled perception) have been linked to ampclusion complaints. These factors have an amplifier origin and can confound one’s diagnosis of the true origin of the AE. To facilitate the diagnosis (or minimizing the occurrence of ampclusion complaint with an amplifier origin), one may select hearing aids that have a short group delay (less than 10 ms) and those that minimize saturation distortion at both the input and output stages of processing. If ampclusion complaints occur, one will have eliminated the possibility of delay and distortion as possible reasons for the complaint.

Conclusions
There are many factors that affect the hollowness or ampclusion complaint.2 If it is assumed that this complaint is simply related to the amount of low frequency energy within the wearer’s ear canal, one can conclude that the occlusion caused by the shell/earmold is the most likely reason for ampclusion complaint in people with a high frequency hearing loss that is less than 30-40 dB HL at 500 Hz. As the hearing loss in the low frequency increases, the contribution of the amplifier gain increases, and the systematic 5-step procedure—such as that described in Kuk & Ludvigsen4—would be desirable to pinpoint the relative contribution of each factor.

 From left: Francis Kuk, PhD, is the director of audiology, and Heidi Peeters, MA, Denise Keenan, MA, and Chi Lau, PhD, are research audiologists at the Widex Office of Research in Clinical Amplification (ORCA) located in Lisle, Ill, which is a division of Widex Hearing Aid Co, Long Island City, NY.

Correspondence can be addressed to Francis Kuk, Widex Office of Research in Clinical Amplification, 2300 Cabot Dr, Ste 415, Lisle, IL 60532; email: [email protected]

References
1. Painton S. Objective measure of low-frequency amplification reduction in canal hearing aids with adaptive circuitry. J Am Acad Audiol. 1993;4(3):152-156.
2. Kuk F, Ludvigsen C. Ampclusion Management 101: Understanding variables. Hearing Review. 2002;9(8):22-32.
3. Sweetow R, Valla A. Effect of electroacoustic parameters on ampclusion in CIC hearing instruments. Hearing Review. 1997;4(9):8-22.
4. Kuk F, Ludvigsen C. Ampclusion management 102: A 5-step protocol. Hearing Review. 2002;9(9):34-43.
5. Revit L. Two techniques for dealing with the occlusion effect. Hear Instrum. 1992;43(12):16-18.
6. Cortez R, Dinulescu N, Skafte K, Olson B, Keenan D, Kuk F. Changing with the times: applying digital technology to hearing aid shell manufacturing. Hearing Review. 2004;11(3):30-38.
7. Dillon H. Hearing aid earmolds, earshells and coupling systems. Hearing Aids. Turramurra, Australia: Boomerang Press; New York: Thieme Medical Publishing; 2001:117-157.
8. Kuk F, Keenan D, Sonne M, Ludvigsen C. Efficacy of an open fitting hearing aid. Hearing Review. 2005;12(2):26-32.
9. Pirzanski C. Diminishing the occlusion effect: Clinician/manufacturer factors. Hear Jour. 1998;51(4):66-78.
10. Kuk F. Perceptual consequences of venting in hearing aids. Brit J Audiol. 1991;25:163-169.
11. Kuk F, Plager A, Pape N. Hollowness perception in noise-reduction hearing aids. J Am Acad Audiol. 1992;3:39-45.