Translating Compression Research Into Clinical Decisions

Introduction

Amplitude compression was proposed as an amplification strategy over 50 years ago, and has been widely used in hearing aids for 15 years (Dreschler, 1992). Nearly 300 research studies have been published on the use of compression. (Simply perform a Pub Med Search (www.ncbi.nlm.nih.gov) on the keywords "hearing aid compression".) Despite the number of studies, clinicians continue to have questions about choosing the optimal compression parameters for patient needs and how to make compression adjustments in response to patient complaints. The confusion is partly because compression is so complex. In digital hearing aids, clinicians can adjust one or more compression thresholds, compression ratio, attack and time release time in each channel, plus expansion parameters, in addition to adjusting gain, output, crossover frequencies and maximum output. Additionally, the "first fit" option in many digital hearing aid systems tends to hide the access to many of the more advanced parameters in the fitting software. No wonder adjusting a hearing aid is not always straightforward!

It is also the case that some research articles are either not readily available to clinicians, or a clinician may have difficulty translating laboratory studies into clinical decisions. In this article, we link recent research on compression is linked with advice for setting and adjusting compression parameters. Hopefully, providing this type of information will help clinicians to better select the appropriate compression characteristics for a given hearing aid and make effective adjustments.

The format of this paper will be to put forth some of the most commonly asked questions about compression and then provide the answers based on the research evidence as it stands today. This paper concludes with a general protocol that should be considered when initially fitting and adjusting hearing aids. While this exercise may not be able to answer questions to every possibly scenario that a clinician may experience, by providing these overall guidelines, the skilled practitioner will have a solid base of information from which to start.

What do we hope to achieve with multichannel wide-dynamic range compression?

For patients, we usually think of multi-channel wide dynamic range compression in terms of real-life benefits, including the ways in which this type of compression:

• Improves audibility across frequency
  
  
• Prevents loudness discomfort
  
  
• Improves sound intelligibility
  
  
• Improves sound comfort
  
  
• Preserves sound quality
  
  
• Reduces the distortion of important speech cues

In the extreme, some of these goals are mutually exclusive. For example, in the case of a severe loss with very reduced dynamic range, a high compression ratio will improve audibility and prevent loudness discomfort, but also might distort important speech cues (Boothroyd, Springer, Smith, & Schulman, 1988).

Are more compression channels better?

Current digital aids offer up to 20 compression channels. More channels offer flexibility in adjusting compression characteristics to accommodate differences in dynamic range across frequency. Calculations by Woods (Woods, Van Tasell, Rickert, & Trine, 2006) suggest that for most hearing losses, speech audibility is maximized with the use of five channels. On the other hand, using more than five channels does not seem to degrade speech recognition as long as compression ratios are low (Crain & Yund, 1995), and more channels might help with other issues, such as reducing feedback or noise.

How should I set compression threshold and compression ratio?

1. Measure patient thresholds, and measure or predict loudness discomfort levels (Punch, Joseph, & Rakerd, 2004).
   
   
2. Set the compression threshold (also called the kneepoint, the threshold kneepoint [TK], and the compression kneepoint [CK]) at a low enough level to activate compression over the speech dynamic range. The default "first fit" compression threshold for most products is between 40 and 50 dB SPL. Although some studies suggest that subjects prefer higher compression thresholds (Barker & Dillon, 1999), those studies used single-channel compression without expansion and thus hearing aid users experienced unwanted amplification of low-level noise.
   
   
3. Measure the response in the ear, using a conversational-level input (60 to 70 dB SPL). Adjust gain so that the REAR (or REIR) is acceptably close to target. You can also measure the hearing aid response in a coupler and use either an individually-measured or standard RECD to predict the real-ear response (Munro & Hatton, 2000). NOAH simulations, although convenient, may under- or over-estimate the actual response (Aarts & Caffee, 2005; Hawkins & Cook, 2003)s
   
   
4. Adjust gain for soft speech (50 dB SPL input) until the measured response meets the target for soft speech. This requires use of a multi-level prescriptive procedure such as NAL-NL1 or DSL[i/o]. Such prescriptions are now built into most real ear measurement systems, as shown in the figure below.
   
   
   
   Figure 1. Example of a multi-target prescriptive procedure. The graph shows gain (in dB) as a function of frequency. The red curve shows the measured REUR for this patient. The dashed lines marked "1" and "2" show the target insertion gain for conversational-level speech (65 dB SPL) and soft speech (50 dB SPL), respectively. The blue and green curves show the measured REIR for the conversational-level and soft inputs for this patient. Match to target is good through 4 kHz. In this case, the match to target for the soft and conversational inputs was achieved with a 3:1 compression ratio and 45 dB SPL compression threshold.
   
   
5. If you do not have access to a multi-target prescription that generates a soft speech target, you can verify that speech is audible by viewing the measured output relative to the hearing threshold, as in the plot in Figure 2. If soft speech is inaudible, increase the gain for soft speech. It is probably not necessary to guarantee audibility for the full dynamic range of soft speech (Studebaker & Sherbecoe, 2002). And, of course, the importance of hearing low-level signals will vary depending on the patient.
   
   
6. Adjust gain for loud speech so that measured output does not exceed the patient's discomfort thresholds.
   
   
7. In many fitting programs, the compression ratio is determined by the relative gain for soft versus loud speech. In other manufacturers' programs, the compression ratio can be adjusted directly. Note, however, that fitting software varies. For example, increasing the compression ratio might increase soft speech gain while maintaining average gain, or decrease average speech gain while maintaining soft speech gain. In both cases, soft speech is comparatively louder, but in the latter case you will need to increase overall gain to maintain the conversational input at the desired output level.

Figure 2. The figure shows REAR (in dB SPL) as a function of frequency. The red line shows the listener's thresholds, ranging from 50 to 75 dB SPL across frequency. The asterisks show the listener's loudness discomfort levels, ranging from 100 to 115 dB SPL across frequency. The light blue dotted line at the bottom of the figure shows the average normal hearing threshold. The shaded blue area shows the range of aided speech intensities for conversational speech. The low end represents the softest speech levels, the blue line about 2/3 of the way up the range represents the average speech levels, and the top end of the shaded area shows the speech peaks. For this patient, conversational speech should be both audible and comfortable after amplification. [Figure courtesy of Audioscan]

When should I increase the compression ratio?

Compression ratios for severe loss will likely need to be higher than for milder losses in order to fit speech within the listener's dynamic range. Compression ratios might also need to be higher if loudness comfort is a priority. Compression ratios will increase if either the gain for soft sounds is increased, or the gain for loud sounds is decreased, while maintaining gain for other sound levels.

When should I decrease the compression ratio?

If intelligibility is a priority, try to keep the compression ratio at 2:1 or less for mild-to-moderate losses. Most studies (Boike & Souza, 2000; Hohmann & Kollmeier, 1995; Hornsby & Ricketts, 2001; Rosengard, Payton, & Braida, 2005; Verschuure, Prinsen, & Dreschler, 1994) agree that intelligibility is maintained below that point of a 2:1 ratio. Complaints of muffled or unclear sound quality can also be addressed by lowering the compression ratio (Boike & Souza, 2000; Jenstad, Van Tasell, & Ewert, 2003; Neuman, Bakke, Hellman, & Levitt, 1994; Rosengard et al., 2005). Acoustically, this happens because the amplitude contrast in speech has been diminished by use of higher compression ratios. The compression ratio should also be decreased if the patient complains that distant sounds are heard more easily than close sounds, or that background sounds are too loud.

An example is shown in Figure 3. Suppose this patient is at a restaurant. (For now, ignore other hearing aid features like directional microphones or digital noise reduction). Let's imagine that his dining companion's speech reaches his hearing aid microphone at a level of 70 dB SPL. Now suppose someone seated at a nearby table is also talking, and the speech of that person (who is further away) reaches the hearing aid microphone at 50 dB SPL. For the distant talker, the output level is 68 dB SPL (gain 18 dB). For the close talker, the output level is 70 dB SPL (gain 0 dB). Although the desired effect of compression—that soft inputs receive more gain than louder inputs—is maintained, the output difference between the distant talker and the close talker is only 3 dB so the patient may perceive those sounds as of similar loudness.



Figure 3. Input output function for a hearing aid set for a compression threshold of 45 dB SPL, and compression ratio of 6:1.

If this patient comes back for a follow-up and complains that the distant speaker was as loud (or nearly as loud) as the close speaker, we might lower the compression ratio (Figure 4). Now the speech of the distant talker receives 18 dB gain, for an output of 68 dB SPL. The speech of the close talker receives 8 dB gain, for an output of 78 dB SPL. Now the output difference between the speech of the distant talker and the close talker is 10 dB, so the patient will perceive the close talker as louder. These compression ratios are higher than would likely be used in the clinic, but the concept is the same: the higher the compression ratio, the more similar the outputs for sounds of different intensity. The lower the compression ratio, the more dissimilar the outputs for sounds of different intensity will be.



Figure 4. Input output function for a hearing aid set for a compression threshold of 45 dB SPL, and compression ratio of 2:1.

How do changes to compression ratio and compression threshold interact?

In some cases, you can change either the compression ratio or the compression threshold to achieve the desired effect. Figure 5 shows an example. This time, instead of reducing the compression ratio, the compression threshold is increased to 65 dB SPL. Now, a soft input (50 dB SPL) receives a gain of 20 dB, for an output of 70 dB SPL. A louder input (70 dB SPL) receives a gain of 16 dB, for an output of 86 dB. If you compare this with Figure 3 (the original setting which generated the complaint that distant sounds were more easily heard) you will see that by increasing the compression threshold, we have increased the difference in output between soft and louder inputs.



Figure 5. Input output function for a hearing aid set for a compression threshold of 65 dB SPL, and compression ratio of 6:1.

If you watch how compression parameters change as you adjust relative gain for soft or loud sounds in the manufacturer's software, you might notice that sometimes a gain change may affect compression ratio, compression threshold, or both. The specific effect(s) of changing gain is determined by each manufacturer's software.

How should I set attack and release time?

There is little consensus about setting compression time constants. In theory, short time constants offer the best audibility, because they maximize gain for soft consonants within a word. Better consonant audibility should translate to better intelligibility, and it does in at least some studies (Gatehouse, Naylor, & Elberling, 2006; Novick, Bentler, Dittberner, & Flamme, 2001). However, a short release time can also distort usable speech cues (Jenstad & Souza, 2005). Additionally, it's clear that listeners prefer longer release times when speech quality and comfort are listening goals (Gatehouse et al., 2006; Hansen, 2002; Neuman, Bakke, Mackersie, Hellman, & Levitt, 1995).

An additional factor is the recent finding (Gatehouse et al., 2006) that adults with lower cognitive abilities have higher speech intelligibility with longer release times. Adults with cognitive dysfunction are likely to be older, so it seems prudent to use long release times with older adults. Some manufacturers' software makes automatic reaction time adjustments based on patient age. It has also been suggested by Sandridge and Newman(Sandridge & Newman, 2006) that measures such as the Mini Mental Status Examination (MMSE)(Folstein, Folstein, & McHugh, 1975) or other similar scales could be used to screen patients for problems with cognitive function. However, the MMSE is not likely to distinguish subtler cognitive deficits; and more comprehensive instruments are impractical for the audiology clinic. There is a need for a clinically feasible cognitive measure in the audiology practice, and some researchers are working towards this end.

When should I increase the release time?

Increase the release time in response to patient complaints about sound quality, "pumping", or that the hearing aid seems to cut in and out (Jenstad et al., 2003). As described above, longer release times should also be used for older listeners, at least until we have a better way to measure cognition in the audiology clinic.

Putting it all together

If we follow evidence-based practice (Cox, 2005), we first identify the problem. Reducing discomfort from dishes clattering requires a different approach than improving clarity for a spouse's voice. Some patients provide more descriptions than others. In the case of patients who are less adept at describing their complaints, fitting tools like real ear measurements can help to clarify problems with fittings.

From there, we need to know where we're going. Because compression parameters interact, multiple "trial and error" adjustments can cancel each other out or make things worse. The manufacturers' first fit is a good starting point, followed by the suggestions below:

• Use an appropriate prescriptive procedure
  
  
• Verify response in real ear to ensure sufficient audibility and loudness comfort; ensure that you use the proper speech or speech-like stimuli to account for the effect of the noise reduction circuitry
  
  
• Adjust programming in response to patient feedback
  
  
• Elicit descriptions from patients: loud vs. soft, high vs. low frequency, to know what channel or part of the input-output function to target
  
  
• Manufacturers' software uses different configurations, so rely on real ear or NOAH plots to understand the effects of compression changes

More research is underway to solve some additional problems. For example, researchers are working for better cognitive tests to identify patients needing specific compression parameters and on speech enhancement techniques that might counteract some of the negative acoustic effects of compression. Hearing aid manufacturers are also doing more research than ever before and providing new and innovative ways to verify the function of the hearing aid. Because the clinician is often between new technologies that are developed and the research that evaluates those technologies, the author hopes that the practical advice provided in this paper will help bridge the gap between the two. We may not be able to build the perfect hearing aid, but careful adjustment can make the best use of the remarkable hearing technology we have.

Acknowledgments

Thanks to Steve Armstrong at Gennum Corporation for use of the compression simulator, and Eric James at the Seattle VA for helpful comments on this manuscript. The author's work was supported by NIDCD, NCRAR, and the Bloedel Hearing Research Center.

References

Aarts, N. L., & Caffee, C. S. (2005). Manufacturer predicted and measured REAR values in adult hearing aid fitting: Accuracy and clinical usefulness. International Journal of Audiology, 44(5), 293-301.

Barker, C., & Dillon, H. (1999). Client preferences for single-channel compression threshold in single-channel wide dynamic range compression hearing aids. Ear and Hearing, 20(2), 127-139.

Boike, K. T., & Souza, P. E. (2000). Effect of compression ratio on speech recognition and speech quality ratings with wide dynamic range compression amplification. Journal of Speech, Language and Hearing Research, 43(2), 456-468.

Boothroyd, A., Springer, N., Smith, L., & Schulman, J. (1988). Amplitude compression and profound hearing loss. Journal of Speech and Hearing Research, 31(3), 362-376.

Cox, R. M. (2005). Evidence-based practice in provision of amplification. Journal of the American Academy of Audiology, 16(7), 419-438.

Crain, T., & Yund, E. W. (1995). The effect of multichannel compression on vowel and stop-consonant discrimination in normal-hearing and hearing-impaired subjects. Ear and Hearing, 16(5), 529-543.

Dreschler, W. A. (1992). Fitting multichannel-compression hearing aids. Audiology, 31(3), 121-131.

Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12(3), 189-198.

Gatehouse, S., Naylor, G., & Elberling, C. (2006). Linear and nonlinear hearing aid fittings--2. Patterns of candidature. International Journal of Audiology, 45(3), 153-171.

Hansen, M. (2002). Effects of multi-channel compression time constants on subjectively perceived sound quality and speech intelligibility. Ear and Hearing, 23(4), 369-380.

Hawkins, D. B., & Cook, J. A. (2003). Hearing aid software predictive gain values: How accurate are they? Hearing Journal, 56(7), 26-34.

Hohmann, V., & Kollmeier, B. (1995). The effect of multichannel dynamic compression on speech intelligibility. Journal of the Acoustical Society of America, 97(2), 1191-1195.

Hornsby, B. W., & Ricketts, T. A. (2001). The effects of compression ratio, signal-to-noise ratio, and level on speech recognition in normal-hearing listeners. Journal of the Acoustical Society of America, 109(6), 2964-2973.

Jenstad, L. M., & Souza, P. E. (2005). Quantifying the effect of compression hearing aid release time on speech acoustics and intelligibility. Journal of Speech, Language and Hearing Research, 48(3), 651-667.

Jenstad, L. M., Van Tasell, D. J., & Ewert, C. (2003). Hearing aid troubleshooting based on patients' descriptions. Journal of the American Academy of Audiology, 14(7), 347-360.

Munro, K. J., & Hatton, N. (2000). Customized acoustic transform functions and their accuracy at predicting real-ear hearing aid performance. Ear and Hearing, 21(1), 59-69.

Neuman, A. C., Bakke, M. H., Hellman, S., & Levitt, H. (1994). Effect of compression ratio in a slow-acting compression hearing aid: paired-comparison judgments of quality. Journal of the Acoustical
Society of America, 96(3), 1471-1478.

Neuman, A. C., Bakke, M. H., Mackersie, C., Hellman, S., & Levitt, H. (1995). Effect of release time in compression hearing aids: paired-comparison judgments of quality. Journal of the Acoustical Society of America, 98(6), 3182-3187.

Novick, M. L., Bentler, R. A., Dittberner, A., & Flamme, G. (2001). Effects of release time and directionality on unilateral and bilateral hearing aid fittings in complex sound fields. Journal of the American Academy of Audiology, 12(10), 534-544.

Punch, J., Joseph, A., & Rakerd, B. (2004). Most comfortable and uncomfortable loudness levels: Six decades of research. American Journal of Audiology, 13, 144-157.

Rosengard, P. S., Payton, K. L., & Braida, L. D. (2005). Effect of slow-acting wide dynamic range compression on measures of intelligibility and ratings of speech quality in simulated-loss listeners. Journal of Speech, Language and Hearing Research, 48(3), 702-714.

Sandridge, S. A., & Newman, C. W. (2006). Improving the efficiency and accountability of the hearing aid selection process: use of the COAT. Audiology Online, Article 1541.

Studebaker, G. A., & Sherbecoe, R. L. (2002). Intensity-importance functions for bandlimited monosyllabic words. Journal of the Acoustical Society of America, 111(3), 1422-1436.

Verschuure, H., Prinsen, T. T., & Dreschler, W. A. (1994). The effects of syllabic compression and frequency shaping on speech intelligibility in hearing impaired people. Ear and Hearing, 15(1), 13-21.

Woods, W. S., Van Tasell, D. J., Rickert, M. A., & Trine, T. D. (2006). SII and fit-to-target analysis of compression system performance as a function of number of compression channels. International Journal of Audiology, 45(11), 630-644.