Applying Expansion in Hearing Aid Fittings: Subjective and Objective Findings

Introduction

One goal of wide-dynamic-range compression (WDRC) hearing instruments is to improve the audibility of low-level, high-frequency speech cues necessary for accurate speech understanding (Johnson, 1993; Killion, 1996). To that end, WDRC hearing instruments provide level-dependent amplification such that low-level input signals receive more amplification than high-level input signals. Input signal levels below the threshold of compression typically receive linear amplification, thereby resulting in the maximum amplification available by the hearing instrument. The provision of maximum amplification for low-level signals is intended to improve speech intelligibility by increasing the audibility of speech cues necessary for feature identification; however, excessive amplification of low-level signals can significantly increase the audibility of low-level environmental noise, as well as noise generated by the hearing instrument itself (Ricketts & Henry, 2002).

Hearing instrument noise typically originates from the microphone (Kuk, 2002) at a level of approximately 20 dB SPL in most modern devices (Holube & Velde, 2000). Internal noise that is amplified by the hearing instrument may be heard by the hearing instrument user (Kuk, 2002), thereby creating the criticism that WDRC hearing instruments are abnormally noisy when used in quiet and/or low-level environments (Ghent, Nilsson, & Bray, 2000; Venema, 1998). In fact, WDRC hearing instrument users commonly note an audible "hissing" sound when listening in quiet and/or low-level environments (Venema, 2000). Furthermore, the amplification of low-level noise may be particularly annoying for listeners with hearing thresholds at or near normal for some frequency regions (Holube & Velde, 2000).

Villchur (1973) observed that excessive amplification of low-level inputs was an undesirable side-effect of compression and suggested that reduced amplification for low-level inputs may alleviate this problem in hearing instruments. Technology designed to reduce the amount of amplification of low-level inputs is known as expansion (Kuk, 2002). Expansion is opposite of WDRC, in that signals below the compression threshold receive reduced gain rather than maximum gain (Venema, 2000). Consequently, low-level environmental noise and internal noise generated by the hearing instrument should be less audible when expansion is activated (Kuk, 2002; Ghent et al., 2000; Venema, 1998).

Although expansion is commonly recommended for WDRC hearing instrument users with near-normal thresholds to alleviate complaints of noise when in low-level environments (Venema, 2000), decreasing the amplification of low-level signals may also reduce the audibility of low-level, high-frequency speech cues necessary for accurate speech recognition. Consequently, expansion may provide improved sound quality at the cost of reduced recognition of low-level speech. Several studies were conducted by the author and colleagues to better quantify this relationship and possibly provide recommendations on how expansion should be used in hearing aid fittings.

Single-Channel Expansion

To determine the extent of the trade-off described above, Plyler, Hill, and Trine (2005) examined the effects of single-channel expansion on the objective and subjective performance of twenty hearing instrument users with varying degrees of hearing loss. The focus of the research was to answer the following questions: 1) does single-channel expansion affect the recognition of low-level speech in quiet and in noise?; 2) are listeners more satisfied with the amount of low-level noise reduction provided by the hearing instrument when using single-channel expansion?; 3) does single-channel expansion affect overall listener preference?; and 4) are the effects of single-channel expansion related to the degree of hearing loss of the listener?

Methods

Twenty participants with sensorineural hearing loss were equally divided into two groups based on hearing sensitivity. One group had two adjacent hearing thresholds better than 40 dB HL from 250 through 1000 Hz while the other group had two adjacent hearing thresholds that were poorer than 40 dB HL from 250 through 1000 Hz. All participants were fitted binaurally with digital in-the-ear hearing instruments with single-channel WDRC processing and multiple memory capability (Starkey Endeavour). Two memories of the hearing instruments were programmed for each participant such that one memory activated single-channel expansion and one memory deactivated single-channel expansion. The expansion kneepoint was set to 50 dB SPL; the expansion attack and release times were 512 ms, and the expansion ratio was 0.5:1. All other fittings parameters were held constant across the two memories, and the noise suppression feature and volume control were deactivated for the entire experiment.

Objective performance was evaluated in quiet using the Connected Speech Test (CST) (Cox, Alexander, & Gilmore, 1987; Cox, Alexander, Gilmore, & Pusakulich, 1988) and in noise using the Hearing in Noise Test (HINT) (Nilsson, Soli, & Sullivan, 1994). Both the CST and HINT were conducted at 40, 50, and 60 dB SPL to ensure speech levels were below, at, and above the expansion threshold. Subjective performance was evaluated by having each participant rate their satisfaction regarding the amount of noise reduction they perceived in each expansion condition (on versus off) on a daily basis while using the hearing instruments in their own homes. At the completion of a two-week trial period, each participant was asked to indicate which expansion condition they preferred overall.

Results

Results indicated that single-channel expansion significantly degraded speech recognition in quiet and in noise when input levels were at or below the threshold of expansion; however, single-channel expansion significantly improved subjective performance for all listeners when used in a quiet environment and was preferred to the no-expansion condition overall. Furthermore, single-channel expansion affected the objective and subjective performance of listeners with varying degrees of hearing loss in a similar manner.

Deficits in speech recognition observed for stimuli presented at or below the expansion threshold were attributed to reduced audibility of the speech signal during the expansion-on condition. Amplitude variations in the speech waveform resulted in speech levels that were, at times, below the expansion threshold, thereby activating the expansion feature and significantly reducing the audibility of low-level speech. An explanation for why participants preferred the expansion-on condition despite the fact that expansion significantly reduced speech recognition ability was less clear. It is possible that some participants did not listen to speech when evaluating expansion while others rarely listened to speech at such low-levels. In either case, speech recognition deficits would have gone unnoticed and would not be expected to impact subjective results. On the other hand, the noise reduction benefits produced by expansion may have simply outweighed any speech recognition deficits produced.

Lastly, the findings from this initial study of single-channel expansion were not in agreement with conventional thinking regarding the use of expansion. As previously mentioned, expansion is typically recommended for listeners with near normal thresholds in order to alleviate complaints of noise when in quiet settings. Plyler et al. (2005) demonstrated, however, that single-channel expansion affects performance and preference similarly, regardless of the degree of hearing loss of the listener. Thus, single-channel expansion provides improved sound quality at the cost of reduced recognition of low-level speech not only to listeners with near normal thresholds but to listeners with varying degrees of hearing loss as well (Plyler et al., 2005).

Multi-Channel Expansion

Although single-channel expansion improved subjective evaluations for WDRC hearing instrument users, despite degrading the recognition of low-level speech (Plyler et al., 2005), expansion effects may differ in hearing instruments with multi-channel expansion. For example, expansion parameters such as expansion threshold and expansion ratio may vary across channels in multi-channel hearing instruments in order to activate expansion in restricted frequency regions as opposed to across the entire spectrum. Therefore, multi-channel expansion may simultaneously preserve audibility of high frequency speech cues necessary for accurate feature identification while reducing gain for low-level, low-frequency environmental noise and noise generated by the hearing instrument. If so, the use of multi-channel expansion could maintain subjective improvements for WDRC hearing instrument users without degrading the recognition of low-level speech.

Consequently, Plyler, Lowery, Hamby and Trine (2007) examined the effects of multi-channel expansion on the objective and subjective evaluation of 20 listeners fitted binaurally with multi- channel, digital, in-the-ear hearing instruments. The goal of the research was to answer the following questions: 1) does multi-channel expansion affect the recognition of low-level speech in quiet and in noise?; 2) does multi-channel expansion affect satisfaction with low-level noise reduction?; and 3) does multi-channel expansion affect overall listener preference?

Methods

Twenty participants with sensorineural hearing loss were fitted binaurally with digital, in-the-ear hearing instruments with four-channel WDRC processing and multiple memory capability (Starkey Axent II). The crossover frequencies between the first and second, second and third, and third and fourth channels of each hearing instrument were 750, 1750, and 3750 Hz respectively. The expansion parameters were constant across all hearing instruments such that each hearing instrument had an expansion threshold of 50 dB for channels one and two, 40 dB for channel three and 30 dB for channel four. These expansion thresholds corresponded to the long-term spectrum of low-level speech plus 5 dB to preserve amplification for soft speech while allowing expansion activation for low-level noise. The expansion time constant was 512 ms and the expansion ratio was 0.5:1 for each channel of each hearing instrument.

Three memories of the hearing instruments were programmed for each participant in a random order such that each hearing instrument had one memory in which expansion was activated in all four channels (full), one memory in which expansion activation was restricted to channels one and two only (restricted), and one memory in which expansion was deactivated in all four channels (off). Expansion activation was restricted to channels one and two to determine if preserving audibility of high-frequency speech cues while reducing amplification of low-level, low frequency environmental/hearing instrument noises would maintain subjective improvements without degrading the recognition of low-level speech. All other fitting parameters were held constant across the three memories, and the noise suppression feature and volume control were deactivated for the entire experiment.

As in the Plyler et al. (2005) study, objective performance was evaluated in quiet using the CST (Cox et al, 1987; Cox et al, 1988) and in noise using the HINT (Nilsson et al., 1994). Both measures were conducted at 40, 50, and 60 dB SPL to allow for direct comparison of single-channel and multi-channel results. Subjective performance was evaluated by having each participant rate their satisfaction regarding the amount of noise reduction they perceived in each expansion condition (full, restricted, off) on a daily basis while using the hearing instruments in their own homes. At the completion of a two-week trial period, each participant was asked to indicate which expansion condition they preferred overall.

Results

Listeners performed significantly better in quiet and in noise when multi-channel expansion was off than when either multi-channel condition was active (full or restricted). However, limiting expansion to channels one and two improved objective performance in quiet and in noise relative to the four-channel condition. On the other hand, satisfaction ratings were significantly greater for both multi-channel conditions (full and restricted) than for the off condition; however, satisfaction ratings were similar for the restricted and the four-channel conditions. Overall, listeners preferred any form of multi-channel expansion to no expansion. Additionally, overall preference was not significantly different between the restricted and the four-channel conditions.

Speech recognition deficits observed in this research were in agreement with previous findings obtained using single-channel expansion (Plyler et al., 2005) and were attributed to reduced audibility of the speech signal during both multi-channel expansion conditions. As expected, multi-channel expansion significantly reduced in-situ levels when the input level of the stimulus was below the expansion threshold. As a result, the audibility of necessary speech cues may have been reduced for both multi-channel expansion conditions (full and restricted), thereby degrading low-level speech recognition in quiet and in noise. That being said, listeners did perform significantly better in quiet and in noise for the restricted condition than the four-channel condition, possibly due to the fact in-situ levels were an average of 10 dB less for the four-channel condition than the restricted condition from 1235 to 2000 Hz. Therefore, restricting expansion to below 2000 Hz reduced, but did not overcome, speech recognition deficits observed when expansion was active across the entire spectrum.

Subjective results indicated that restricting expansion to channels one and two did not significantly affect satisfaction ratings or overall preference relative to the four-channel condition. A comparison between in-situ levels obtained for the restricted and the four-channel conditions revealed similar response levels from 500 to 1235 Hz; however, in-situ levels were an average of 10 dB less for the four-channel condition than the restricted condition from 1235 to 6000 Hz. Reducing audibility above 1235 Hz for the four-channel condition, however, did not further improve subjective performance relative to the restricted condition. Therefore, subjective benefit from multi-channel expansion was attributed to expansion activation below 1235 Hz.

As in single-channel research, however, findings demonstrated that hearing instrument users favor the use of multi-channel expansion even though multi-channel expansion significantly degraded the ability to recognize low-level speech in quiet and in noise. As in previous research, participants evaluated each expansion condition in their own homes instead of in the laboratory in order to examine the effects of multi-channel expansion in the real-world settings for which expansion was designed. Thus, it is possible that ambient levels did not fall below the threshold and engage expansion during some subjective evaluations or that participants were not actively listening to speech when evaluating multi-channel expansion. Nonetheless, objective and subjective findings from this multi-channel expansion experiment were in agreement with previous conclusions from single-channel expansion research that suggested single-channel expansion improves sound quality at the cost of reduced recognition of low-level speech (Plyler et al., 2005).

Low-Frequency Expansion

Lowery and Plyler (in press) extended the work of Plyler et al. (2007) to determine the effects of low-frequency (
The goal of the research was to answer the following: 1) does low-frequency expansion affect the recognition of low-level speech in quiet and in noise?; 2) does low-frequency expansion affect satisfaction?; and 3) does low-frequency expansion affect overall listener preference?

Methods

Each hearing instrument used was the same digital, in-the-ear device with four-channel WDRC processing and multiple memory capability (Starkey Axent II) used in the Plyler et al. (2007) experiment. Three memories of each hearing instrument were programmed in a random order such that expansion was activated in channel one only (
As in the previous studies discussed, objective performance was evaluated using the CST and HINT; however, both measures were conducted at 40 dB SPL only. In previous expansion studies, subjective evaluations were conducted in the participants' own homes instead of in the laboratory to examine expansion in the real-world settings for which it was designed (Plyler et al., 2005; 2007). In this experiment, subjective evaluations were conducted in a laboratory to determine if laboratory findings agree with those obtained in more real-world listening environments. Furthermore, participants were seated in a sound-treated examination room and rated their satisfaction for each memory while listening in quiet and while listening to CST sentences (Cox et al., 1987) presented at 40 dB SPL. The presentation level of 40 dB SPL ensured the activation of expansion for those memories in which expansion could be activated. At the completion of the experiment, each participant was asked to indicate which setting they preferred overall to determine if low-frequency expansion affected overall preference.

Results

Results indicated that speech recognition in quiet and in noise was similar for the off and channel one conditions; however, listeners performed significantly better in quiet and in noise for the off and channel one conditions than for the channels one-and-two condition. Satisfaction ratings revealed a significant interaction between expansion and listening environment. Satisfaction ratings were significantly higher for both expansion conditions than the off condition when listening in quiet; however, satisfaction ratings were significantly higher for the off and channel one conditions than the channels one-and-two condition when listening to low-level speech. Overall, listeners preferred expansion in channel one only to expansion in channels one and two; however, overall preference was similar for the channel one and off conditions.

Objective findings from this study suggested that limiting expansion activation to 1000 Hz and below did not result in significant speech-recognition deficits observed when expansion was active across a broader spectrum, possibly due to improved audibility of necessary speech cues between 1000 and 2000 Hz. Furthermore, subjective findings obtained in the laboratory were in agreement with previous findings obtained using multi-channel expansion in real-world settings, in that multi-channel expansion significantly improved satisfaction ratings in quiet relative to the no expansion condition, presumably due to reduced audibility of low-level environmental noises as well as noises generated by the hearing instrument (Plyler et al., 2007). However, results obtained when listening to low-level speech in the laboratory were not in agreement with previous findings obtained using either single-channel or multi-channel expansion in real-world settings which indicated that any form of expansion significantly improved listener satisfaction ratings despite the fact expansion significantly reduced low-level speech-recognition (Plyler et al., 2005; 2007).

Subjective outcome differences observed between the present and previous research were attributed to the methodological differences used between investigations. In previous research, participants evaluated each expansion condition in their own homes instead of in the laboratory to evaluate expansion in the real-world settings for which it was designed (Plyler et al., 2005; 2007). In the Lowery and Plyler study (2007), subjective evaluations were conducted in the laboratory to determine if laboratory findings agreed with those obtained in more real-world environments. This design ensured that ambient levels fell below the expansion thresholds and engaged the feature during quiet evaluations and that participants actually listened to low-level speech when evaluating each expansion condition. Interestingly, satisfaction ratings results obtained when listening to low-level speech were in agreement with speech recognition results in that performance for both measures was significantly reduced when expansion was activated in channels one and two while performance improved significantly when expansion was limited to channel one or was deactivated. Furthermore, preference results were also in agreement with low-level speech satisfaction ratings and objective results in that 93% of the participants preferred deactivating expansion or limiting expansion activation to channel one, conditions which maximized speech recognition in quiet, speech recognition in noise, and satisfaction ratings when listening to low-level speech. Therefore, limiting expansion activation to 1000 Hz and below appears to preserve both objective and subjective performance in quiet and in low-level environments, and appears to be preferred overall by a significant number of participants.

Conclusion

The purpose of this paper was to summarize recent research findings regarding the use of expansion in single- and multi-channel WDRC hearing instruments. Dispensers should be aware that expansion may improve subjective outcome at the cost of low-level speech recognition. Research suggests that the deleterious effects of expansion may be overcome by limiting expansion activation to frequency regions below 1000 Hz.

Dispensers should also be aware that subjective performance with expansion depends on whether the end user is listening in a quiet environment or is listening to low-level speech. Lastly, dispensers should pay careful consideration to the specific expansion parameters used in the hearing instruments they dispense. The effect of expansion on low-level speech and noise will depend on the expansion parameters in each device; therefore, findings presented here may not generalize to other types of expansion systems (e.g., more than four channels or other expansion parameters such as threshold, ratio, and time constant). Future work should investigate the role these parameters play on the effectiveness of expansion.

References

Cox, R., Alexander, G., & Gilmore, C. (1987). Development of the Connected Speech Test (CST). Ear and Hearing, 8, 119-126.

Cox, R., Alexander, G., Gilmore, C., & Pusakulich, K. (1988). Use of the Connected Speech Test (CST) with hearing-impaired listeners. Ear and Hearing, 9, 198-207.

Ghent, R.M., Nilsson, M.J., & Bray, V.H. (2000). Uses of expansion to promote listening comfort with hearing aids. Poster presented at the American Academy of Audiology Convention. Chicago, Illinois.

Johnson, W.A. (1993). Beyond AGC-O and AGC-I: Thoughts on a new default standard amplifier. The Hearing Journal, 46, 63-69.

Holube, I., & Velde, T.M. (2000). DSP hearing instruments. In R.E. Sandlin (Ed.), Textbook of Hearing Aid Amplification. San Diego: Singular Publishing Group.

Killion, M.C. (1996). Compression distinctions. The Hearing Review, 3, 29-32.

Kuk, F.K. (2002). Considerations in modern multi-channel nonlinear hearing aids. In M. Valente, M (Ed.), Hearing Aids: Standards, Options, and Limitations. New York: Thieme Medical Publishers.

Lowery, K., & Plyler, P.N. (in press). The objective and subjective evaluation of low-frequency expansion in wide dynamic range hearing instruments. Journal of the American Academy of Audiology.

Nilsson, M., Soli, S., & Sullivan, J. (1994) Development of the Hearing in Noise Test for the measurement of speech reception threshold in quiet and in noise. Journal of the Acoustical Society of America, 95, 1085-1099.

Plyler, P.N., Hill, A.B., & Trine, T.D. (2005). The effects of expansion on the objective and subjective performance of hearing instrument users. Journal of the American Academy of Audiology, 16,101-113.

Plyler, P.N., Lowery, K., Hamby, H., & Trine, T.D. (2007). The effects of multi-channel expansion on the objective and subjective performance of hearing instrument users. Journal of Speech, Language, and Hearing Research, 50, 15-24.

Ricketts, T. & Henry, P. (2002). Low-frequency gain compensation in directional hearing aids. American Journal of Audiology, 11, 29-41.

Venema, T.H. (1998). Programmable and digital hearing aids. In T.H. Venema (Ed.), Compression for Clinicians. San Diego: Singular Publishing Group Inc.

Venema, T.H. (2000). The many faces of compression. In R.E. Sandlin (Ed.), Textbook of Hearing Aid Amplification. San Diego: Singular Publishing Group.

Villchur, E. (1973). Signal processing to improve speech intelligibility in perceptual deafness. Journal of the Acoustical Society of America, 53, 1646-1657.