As hearing aid technology advances, the hearing instrument is evolving more and more into a “hearing system.” The main objective of complex digital signal processing in a hearing system is to improve speech intelligibility in any acoustic environment. Besides that, improvements in other hearing aid features have also been developed. For example, the modern hearing system should provide speech intelligibility with a pleasant tone and at a comfortable level. Cosmetic appearance should be high while operation of the instrument should be easy.
The hearing healthcare professional is constantly being given more tools for (interactive) fitting and fine-tuning of these complex hearing systems.
Because of the high cost of hearing aid development, it is imperative to measure the efficiency of newly developed features and their benefit to the user. This paper will review the results of a clinical study of 20 subjects who tested two new features of the Phonak Claro 22 digital hearing instrument: the Loudness Perception Profile, an in situ categorical loudness scaling procedure designed to obtain an individual hearing aid pre-fitting based on measures, not estimates, of loudness recruitment, and the AutoSelect mode, which switches automatically between two programs to select the better one for the acoustical situation.
TWO PROGRAMMING OPTIONS
Initial programming of the test instruments can be based on either the pure-tone audiogram alone or supplemented by the Loudness Perception Profile (LPP). The LPP is a fine-tuning tool for the programming of individual signal processing parameters of the hearing instrument. Based on categorical loudness scaling,1,2 it takes into account the individual's measured, not estimated, loudness recruitment.
To establish the LPP, the hearing instrument is made to generate narrow-band test signals of varying frequency and loudness. The wearer ranks the perceived loudness of the test signals on a scale of categories from “very loud,” “loud,” “medium,” “soft” to “very soft” (including four intermediary steps), as well as the categories “inaudible” and “extremely loud” at the extremes of the scale. The level of the test signals is adapted throughout the test.
The AutoSelect mode of the test instrument permits automatic switching between Hearing Programs 1 and 2, depending on the acoustic environment. For quiet situations without background noise, AutoSelect will activate Program 1, which is designed to optimize speech intelligibility and sound quality in such situations. For louder, noise-affected situations, AutoSelect will activate Program 2, in which a high-resolution noise-reduction system seeks to optimize the signal-to-noise ratio and comfort for communication in background noise. The classification algorithm implemented in the test instruments is based on a four-dimensional analysis proposed by Kates3 and monitors the overall loudness, fluctuations in overall loudness, spectral center, and fluctuations in the spectral center. The test instrument switches between the programs when all four criteria are met in any given acoustic environment. Switching takes a minimum of 10 seconds. The objective of the implemented noise classification is to differentiate speech within a broad-band noise signal from all other acoustic signals.
A CLINICAL STUDY
Twenty hearing-impaired subjects—9 women and 11 men—participated in this study. They ranged in age from 32 to 81 years with a mean age of 60 (SD +/−15) years. All subjects had a moderate, symmetrical sensorineural hearing loss, had worn hearing aids for many years, and were satisfied with the way the hearing aids were set. Figure 1 shows the thresholds and uncomfortable loudness levels (UCL) averaged over the 20 subjects, as well as the standard deviations.
Schedule and measurement procedures
Each participant was fitted binaurally and the test hearing aids were pre-programmed using the audiogram and then again using the LPP. Each participant tested the two settings, one after the other. Ten subjects judged pre-setting by audiogram first, while the 10 other subjects first judged the setting using the LPP.
The initial setting of the hearing instruments was evaluated for different acoustic situations by means of an “environment tour” in the surroundings of the Hörzentrum Oldenburg. Each subject was accompanied by the senior researcher while evaluating the hearing instrument settings in the open, in a café, in quiet, and while listening to music. With the help of a questionnaire, the subject reported on the subjectively experienced loudness, speech intelligibility, and sound quality for several sound dimensions, as well as the overall impression at the end of the tour. After the subjects gained an initial impression of both settings in different acoustic environments, they were asked for their preferred pre-setting. Their choice was programmed into the test instruments and was tested for 1 week in everyday life at home. The instrument was fine-tuned at the next appointment on the basis of the preferred setting.
The intelligibility of single words in quiet for the fine-tuned test instruments was measured using a rhyme test.4 A list of 47 monosyllabic words was presented in the free field via a loudspeaker at a level 50 dB SPL. For each word there were five alternate answers, which were always differentiated by one phoneme.
The speech intelligibility in background noise was tested using the adaptive Göttinger Sentence Test.5 A list of 20 sentences was presented to the test subjects via a loudspeaker in the free field. Simultaneously, a continuous noise of 65 dB SPL was presented. Both signals were presented from 0° azimuth. Car noise was used as a stationary noise sample and the modulated ICRA-76 noise as a fluctuating sample. The volume of the speech was regulated for each subject so that the sentence test yielded a score of L50. L50 is the speech-to-noise ratio (SNR) at which speech intelligibility is 50%.
To ensure that the subjects were able to evaluate the functionality of the Auto Select mode in everyday life confidently, they were asked to describe and compare the first and second programs in up to four different typical personal listening environments in the first week after the fine tuning. Each subject had access to a remote control in order to switch between the two programs manually. The Auto-Select mode was deactivated during this phase. The participants were carefully instructed, as the aim of this training phase was to teach them to differentiate confidently between the programs. To aid them in their descriptions, they were given volume, sound quality, speech intelligibility, the relationship of speech level to noise level, and listening comfort as points of reference.
When the subjects had become familiar with both programs and been trained to recognize them, the hearing aids were set to automatically switch on in Auto Select mode in the following 6 weeks. During this longer test phase, subjects were asked which program was chosen by the AutoSelect for those situations that they had described in the training phase. While the subjects wore the test instruments at home, they answered a questionnaire developed by Phonak which incorporates, among other things, part of the Oldenburg Inventory.7 The questionnaire included questions about loudness, speech intelligibility, sound quality, and the handling of the instruments—factors that are independent of the hearing aid type. It also contained questions specific to the test instrument, e.g., about the functionality of the Auto Select mode.
The data collected for speech intelligibility, sound quality, loudness, and overall impression for each setting in the “environment tour” are presented in Figure 2. The settings based on scaling resulted in the loudness being rated as comfortable, whereas the loudness was rated a bit too loud for the audiogram-based setting. This difference in loudness was statistically significant (t-test, p<0.05) for outdoor sounds and in quiet environments. There was no statistically significant difference between the settings with regard to subjective speech intelligibility. The sound quality of music was rated statistically significantly better for the setting based on scaling.
Nine subjects preferred the LPP-based setting after the environment tour (Subgroup LPP), and 11 subjects preferred the audiogram-based settings (Subgroup A).
The individually preferred initial settings provided the basis for fine-tuning Program 1. The changes made in Program 1 were automatically carried through to Program 2 by the software. Figure 3 shows the frequency of change of different hearing aid parameters during the fine-tuning procedure for both Groups A and LPP.
The results of the speech tests obtained with the fine-tuned setting of the test instruments are shown in Table 1 for the groups A and LPP. No statistically significant differences were found between the results of the speech tests for the two groups (t-test, p<0.05).
The subjects tested and reported on the functionality and reliability of the AutoSelect mode in everyday life. The notes of 17 subjects were suitable for evaluation. The subjects had described a total of 20 situations in quiet environments, e.g., in the house, watching TV, at lectures, at musical performances. The results indicated that the AutoSelect switched to Program 1 80% of the time in these situations and to Program 2 only 10% of the time. For 10% of these situations, the subjects could not discern which program was selected.
Altogether, 44 noisy situations were described, e.g., shopping at the supermarket, events with many people, driving. For these situations, the subjects reported a 61% rate of switching to Program 2 by the AutoSelect mode, and a 23% rate to Program 1. Sixteen percent of the time the subjects were unsure which program had been chosen by the AutoSelect.
In addition to the individual acoustic environments at home, the subjects also reported on the AutoSelect mode with the help of a questionnaire. Figure 4 shows how often for specifically described acoustic situations Program 1 or 2 was chosen by the AutoSelect mode in the opinion of the subjects. In quiet situations the Auto-Select chose Program 1 75% to 90% of the time, while in noisy situations Program 2 activated automatically 55% to 75% of the time.
Furthermore, the subjects indicated that they had chosen the Auto-Select mode 70% of the time and had used the remote control to switch between programs the rest of the time. Asked how often the test instrument had switched into the program that they had expected, 45% of the subjects answered “often” and 35% answered “sometimes” on a scale of “never,” “seldom,” “sometimes,” “often,” or “always.” The other 20% of the subjects chose one of the other three categories.
Asked if the automatic program choice was suited to the situation, 15% of the subjects, using the same scale, answered “always,” 40% said “often,” 15% said “sometimes,” and 20% said “seldom.” Asked to rate the AutoSelect mode on a scale of “very little use,” “little use,” “some use,” “quite useful,” or “very useful,” 20% of the subjects perceived it as of “little use,” whereas 40% of the subjects described it as “very useful” and 40% as “quite useful.”
The subjective perception of loudness for the initial setting of instruments based on the LPP was immediately described as “comfortable,” whereas the audiogram-based settings were found to be a little loud. Although the overall gain had to be modified for both settings, for those subjects whose settings were based on the LPP fewer adjustments to the frequency response were necessary at the fine tuning compared to those whose settings were based on the audiogram.
The subjects were divided almost equally on whether they preferred initial settings based on LPP or on the audiogram. They judged both strategies as essentially good and pleasant. When asked the reason for their choice, the subjects cited loudness and pleasantness of sound.
The preferred settings, however, depended more on their individual tastes. An analysis of hearing loss of the subjects who preferred the LPP setting showed they had a lower average hearing loss and lower average UCLs than did those who preferred the audiogram setting. On the whole, the subjects welcomed the interactive nature of setting instruments with the LPP, especially those who suffered from tinnitus and/or were sensitive to loud sounds. Since the subjects with each type of fitting also had good speech-intelligibility scores in quiet and noisy environments, the LPP provides an alternate strategy to the programming of hearing instruments with the audiogram.
The subjects used AutoSelect about 2/3 of the time. None of them reported that the AutoSelect continually switched wrongly or that it caused annoyance. To ascertain the functionality of the Auto-Select mode this study relied wholly on subjective reports. In the “real-life” questionnaire, the subjects made written notes for a number of given environments, and some for individual situations of their home life, as to which program they thought the hearing instrument had switched to.
The two sets of data were in agreement. The analysis of the questionnaire shows that the AutoSelect mode is generally (in about 75% of cases) able to differentiate noisy situations from quiet environments and to switch to the better program for a given situation. For the situation “dialogue with a person in a noisy environment,” the instrument switched into Program 1 according to half the subjects and into Program 2 according to the other half. In this situation, the choice of program seemed to be particularly sensitive to the required acoustic parameters. It would be useful to measure the acoustic parameters and to determine objectively the choice of program in a further study in order to improve the criteria for switching. But, even if the AutoSelect mode was able to categorize any given acoustic environment 100% correctly, it would still be unable to anticipate the individual preferences of a particular hearing aid wearer. For such situations, the option of manual program choice was particularly helpful to the wearers.
In summary, this study has shown that the LPP is an alternate, useful tool for programming the test instruments, which takes into account the individual needs of the hearing-impaired person. The automatic program switching, AutoSelect, was able, in the majority of acoustic environments, to switch to the optimal program and proved to be a useful and user-friendly hearing instrument feature in everyday life situations.