INTRODUCTION: HEARING AID FITTING AND VERIFICATION
Modern multi-channel hearing aids have many adjustable parameters. It is usually possible to adjust their compression threshold and the gain at two or more levels, either separately in each frequency channel or in groups of channels. It may also be possible to adjust the crossover frequency between channels (especially in aids with a small number of channels).
Sometimes it is also possible to adjust the attack and release times of the compression or to select among different forms of compression, such as dual-time constant1,2or fast-acting (syllabic).3,4
Finally, it may be possible to adjust extra features, such as low-level expansion (des-cribed below), feedback cancellation, or noise reduction. The large number of parameters means that it is essential to have some well-defined way of setting the hearing aid to suit the individual patient and to verify that the hearing aid is performing as desired.
Fitting modern hearing aids is typically a two-stage process. The first stage in-volves setting the parameters of the hearing aid according to a formula or computer- based procedure, usually using information from the audiogram, which may be combined with information on loudness discomfort levels. Examples of procedures of this type are DSL(i/o),5 NAL-NL1,6 CAMEQ,7 and CAMREST.8 These procedures give only an initial fit; they are intended to give reasonable gains “on average,” but further adjustment is nearly always needed.
The second stage involves fine-tuning to suit the acoustic and auditory characteristics of individual ears and to suit individual needs and preferences.
It is well established that the gains programmed into a hearing aid do not correspond accurately to the gains achieved for an individual real ear.9,10 Therefore, it is essential to assess the gains actually achieved and to make adjustments when necessary to ensure that the goals of the fitting procedure are met and that signals such as speech are audible over a wide frequency range. This is often done using real-ear measurements based on a probe microphone.11 Typically, the measurements are made using artificial test signals, such as swept pure tones, bands of noise, or a speech-shaped noise (steady noise with a spectral shape that matches the long-term average spectrum of speech).
PROBLEMS WITH THE TRADITIONAL APPROACH
There are several limitations to the traditional way of making real-ear measurements:
❖ The gains actually achieved for real-life signals such as speech and music may differ considerably from the gains measured with steady signals, such as tones and noise. The difference depends on the number of channels in the hearing aid, the speed of the compressors, and the compression thresholds.12-16 This is the case even when features such as noise reduction or feedback cancellation are not present or are not activated.
❖ When a hearing aid in-corporates feedback cancellation, pure-tone test signals may be interpreted by the aid as feedback, and a pure-tone test signal is then partially or completely canceled. The gains measured when this happens may be totally unrepresentative of the gains achieved in everyday life. In some hearing aids, it is possible to disable the feedback-cancellation system, but this may change the effective frequency response of the aid and may also limit the gain that can be achieved.
❖ Many hearing aids incorporate some form of noise reduction. If any particular spectral region appears to be dominated by noise (or by any steady sound), then the gain of the hearing aid in that frequency region is reduced. If the test signal used to assess the gain of the hearing aid is a steady noise or a tone, the gain that is measured may be much less than actually achieved for everyday sounds such as speech and music. In some hearing aids, it is possible to disable the noise-reduction system, but this may change the effective frequency response of the aid, and the gain applied by the aid may differ from that obtained when the noise reduction is active.
❖ The sounds that are used to make the measurements have no relevance to the sounds that the hearing aid user experiences in everyday life, for example the voice of a spouse or parent.
A SOLUTION: SPEECH MAPPING
A solution to the problems described above is provided by “speech mapping” systems such as Aurical Visible Speech from GN Otometrics, Audioscan's Verifit, DSM-PRO from MedRx, Interacoustics Affinity, Visible Speech Mapping from Madsen, and Frye Electronics Live Speech Testing. Such systems can provide a visual display of the short-term spectrum of sounds in the ear canal of the hearing aid user, and provide a direct indication of the audibility of sounds that are important for the user, such as the speech of a partner or other relative.
Speech mapping systems share common features, although they differ in the details of the implementation. I describe here, as an example, the Aurical system. This includes the following hardware (wireless SpeechLink 100 binaural neckset) and the Visible Speech software:
❖ Two modules, each containing a reference microphone and a flexible probe microphone. A module is hung from each ear, using soft and flexible rubber. The probe microphone is initially calibrated relative to the reference microphone, and can then be inserted into the ear canal of the user prior to the hearing aid fitting. The spectra of the signals from the microphones are calculated in real time, both for the reference and probe microphones.
❖ The spectra of the microphone signals are transmitted to a computer (PC) via a wireless link mounted on a flexible rubber “collar.” This collar starts at the lower base of the back of the neck and runs over the shoulders. It is light and comfortable to wear. Because the system is wireless, the client can move around while wearing the system, with the results displayed on screen. This can allow the assessment of speech audibility when the user is listening to a person from the opposite side of the room or even in the next room.
❖ The PC produces a real-time running display of the spectrum in the ear canal. The peak levels can be “held” if desired. The spectrum can be displayed either in a high-resolution format or in a more familiar format based on 1/3-octave bands. In either case, the levels are adjusted so as to reflect the audibility of the ear canal signals at each frequency. The spectral levels can be displayed either in dB SPL or dB HL. The display in-cludes a plot of the client's hearing thresholds and the client's loudness discomfort levels, which can be estimated from the audiogram if they have not been directly measured. The display can also include the “speech banana” showing the range of frequencies and levels occurring in speech at a typical conversational level. The display can be set to “On top” mode, which means that it remains visible on the PC screen even if other software, such as hearing aid fitting software, is running.
❖ If desired, the system can be used with a NOAHlink unit, so that the setting of the hearing aid can be adjusted while the Visible Speech system is in place and running and without any wires connecting the client to the PC. The NOAHlink unit attaches easily to the front part of the collar described above.
ADVANTAGES OF SPEECH MAPPING
A major advantage of the speech mapping approach is that the effective amplification provided by the hearing aid can be assessed using realistic signals such as speech or music and with the aid in its normal mode of operation (with features such as feedback cancellation and noise reduction enabled, if appropriate). Thus, the influence of factors such as number and bandwidth of channels, compression speed, etc., is automatically taken into account.
Furthermore, any effects of feedback cancellation or noise reduction on the performance of the aid are automatically included in the display. Effectively, what you see is what you get. The uncertainties and errors produced by the use of artificial test signals are avoided.
The test signals used to assess hearing aid performance can be generated “live” (for example, by having a spouse or parent talk in a normal conversational way), or sounds can be generated via the sound system of the PC. The software of most speech mapping systems includes a variety of calibrated test signals, including speech in quiet, speech in various types of background sounds, traffic noise, and various types of music. Thus it is possible to simulate many of the everyday situations experienced by the individual client.
Because the displays in speech mapping systems incorporate the audiogram of the client, it is easy to see which parts of the spectrum of the signal, for example speech, are audible to the client. If the peaks of normal conversational speech reach the top of the speech banana over the whole frequency range of the banana, then that speech should be fully audible. The dispenser can thus be confident that the client's ability to understand speech will not be compromised by limited audibility.
If the peaks of normal conversational speech do not reach the top of the speech banana for some frequencies, this indicates that speech audibility is not optimal for those frequencies, and the gain programmed into the hearing aid should be increased appropriately.
If the peaks of normal conversational speech fall well above the top of the speech banana for some frequencies, this may indicate that the hearing aid is providing too much amplification for those frequencies, and it may need to be reprogrammed to reduce the gain.
Some other advantages of speech mapping are listed below:
❖ The display may make apparent effects that would not be revealed by conventional measurements, such as effects of low-level expansion (a reduction of gain for low-level inputs, sometimes called “squelch”). Such low-level expansion is often used in hearing aids to stop the user from hearing low-level noise generated in the aid, typically by the microphone and analog-to-digital converter. Traditional probe-microphone measurements are typically made using test signals with levels ranging from about 50 to 85 dB SPL. Such measurements do not reveal the action of low-level expansion. In many hearing aids the “threshold” below which expansion operates can be programmed, and it is often set to a value between 35 and 45 dB SPL. It is not generally realized that, for weak speech with an overall level of, say, 55 dB SPL, the effective level of the speech in the higher-frequency channels of a hearing aid may be only 30–40 dB SPL.7 If the expansion threshold is above this range, then expansion may be applied at high frequencies, leading to reduced audibility and reduced intelligibility.17 Speech mapping can immediately reveal such an effect, if it is happening. If the client needs to understand low-level speech, then the threshold for the low-level expansion may be reduced or disabled altogether.
❖ The client and his relatives become more involved in the fitting procedure. If a hearing aid is providing only limited audibility of speech, this limitation can immediately be demonstrated and intuitively understood. This is especially important when fitting hearing aids to infants and children. If it is impossible to provide full audibility of speech, perhaps because of the severity of the hearing loss at high frequencies, the client and relatives can understand why this is so and can appreciate the consequences. If a more satisfactory fit can be achieved, the client and relatives can immediately see the difference on screen, and hopefully the client can also hear the difference.
❖ A speech mapping system can indicate whether or not problems with loudness discomfort will occur in everyday life. The pre-recorded test signals can be used to simulate situations where discomfort might be encountered, and the display will indicate if any of the signals approach the loudness discomfort level and, if so, at what frequencies. If necessary, the parameters of the compression or output limiting in the aid can be adjusted to avoid loudness discomfort. In this way, return visits for re-adjustment of the aid(s) are likely to be minimized.18
❖ In fully binaural systems, such as the Aurical Visible Speech system described earlier, it is possible to display the live spectrum of the sound at both eardrums simultaneously. This can be useful for illustrating to the client that sometimes one ear receives a better signal than the other (for example, when a talker is not in front, but to the right or left), and that the use of two hearing aids will prevent signals from the side from being missed. An effective way of illustrating this to the client is to leave the binaural display running throughout the counseling session, and to move around while talking to the client.
Speech mapping provides a valuable tool for the hearing aid dispenser. It allows markedly improved accuracy in the verification and fitting of hearing aids. It provides an immediate indication of the audibility of important everyday signals such as speech, including the speech of relatives. Speech mapping permits the dispenser to adjust the parameters of a hearing aid to optimize speech audibility while avoiding loudness discomfort. It also involves clients and their relatives in the fitting process, leading to greater understanding and satisfaction. Finally, its use is likely to reduce the number of post-fiting visits, thereby saving time and money.
1. Moore BCJ, Glasberg BR: A comparison of four methods of implementing automatic gain control (AGC) in hearing aids. Brit J Audiol
2. Stone MA, Moore BCJ, Alcántara JI, Glasberg BR: Comparison of different forms of compression using wearable digital hearing aids. J Acoust Soc Am
3. Villchur E: Signal processing to improve speech intelligibility in perceptive deafness. J Acoust Soc Am
4. Moore BCJ, Johnson JS, Clark TM, Pluvinage V: Evaluation of a dual-channel full dynamic range compression system for people with sensorineural hearing loss. Ear Hear
5. Cornelisse LE, Seewald RC, Jamieson DG: The input/output formula: A theoretical approach to the fitting of personal amplification devices. J Acoust Soc Am
6. Byrne D, Dillon H, Ching T, et al.: NAL-NL1 procedure for fitting nonlinear hearing aids: Characteristics and comparisons with other procedures. JAAA
7. Moore BCJ, Glasberg BR, Stone MA: Use of a loudness model for hearing aid fitting. III. A general method for deriving initial fittings for hearing aids with multi-channel compression. Brit J Audiol
8. Moore BCJ: Use of a loudness model for hearing aid fitting. IV. Fitting hearing aids with multi-channel compression so as to restore “normal” loudness for speech at different levels. Brit J Audiol
9. Swan IR, Gatehouse S: The value of routine in-the-ear measurement of hearing aid gain. Brit J Audiol
10. Aarts NL, Caffee CS: Manufacturer predicted and measured REUR values in adult hearing aid fitting: Accuracy and clinical usefulness. IJA
11. Mueller HG, Hawkins DB, Northern JL: Probe Microphone Measurements: Hearing Aid Selection and Assessment
. San Diego: Singular, 1992.
12. Stone, MA, Moore BCJ: Syllabic compression: Effective compression ratios for signals modulated at different rates. Brit J Audiol
13. Verschuure J, Maas AJJ, Stikvoort E, et al.: Compression and its effect on the speech signal. Ear Hear
14. Souza PE: Effects of compression on speech acoustics, intelligibility, and sound quality. Trends Amplif
15. Henning RW, Bentler R: Compression-dependent differences in hearing aid gain between speech and nonspeech input signals. Ear Hear
16. Jenstad LM, Souza PE: Quantifying the effect of compression hearing aid release time on speech acoustics and intelligibility. J Sp Lang Hear Res
17. Moore, BCJ. Stainsby TH, Alcántara JL, Kühnel V: The effect on speech intelligibility of varying compression time constants in a digital hearing aid. IJA
18. Cunningham DR, Laó-Dávila RG, Eisenmenger BA, Lazich RW: Study finds use of Live Speech Mapping reduces follow-up visits and saves money. Hear J