Keywords
multichannel hearing aid - channel-free hearing aid - speech in noise - spectral resolution
- temporal resolution
There are many problems associated with hearing loss, so simple sound amplification
is insufficient.[1] Because hearing loss reduces speech intelligibility (SI), as well as audibility,
modern hearing aids contain different signal processing algorithms.[2] Multichannel and channel-free hearing aids have similar goals but differ significantly
in terms of signal processing.
Multichannel hearing aids separate the incoming signal into different frequency bands,
and each band of the signal passes through different amplification channels. In multichannel
compression, each channel contains its compression.[3] Multiple channels in the hearing aid allow the signal level to be estimated and
independent control of gain over relatively narrow frequency ranges.[4] The time constants used in these hearing aids may force the user to choose between
SI (fast-acting system) and listening comfort (slow-acting system).[5]
In light of concerns related to multichannel hearing aids (i.e., channel interaction,
channel summation, spectral smearing, and altered temporal information), channel-free
hearing aids have been developed. The channel-free hearing aid processes the wideband
signal 20,000 times per second. At a basic level, it continuously adjusts the gain
of the hearing aid to amplify each phoneme separately, while maintaining sound quality.
Also, it does this without splitting the signal into fixed channels or bands. It performs
parallel processing by measuring the sound pressure level of the input signal and
assigns a gain to the fed signal into the filter control at any moment as dictated
by the measured sound pressure level. Finally, it amplifies the soft input signal
and maintains the comfort sound for high-level sound without changing the temporal
speech envelope.[6]
[7] While both signal processing strategies are currently available in hearing aids,
it is unclear whether differences in these signal processing strategies affect user
performance and preference.
The aim of our study is to compare the effects of multichannel and channel-free signal
processing techniques used in hearing aids on spectral and temporal resolution, which
is important in speech perception. Nevertheless, it is aimed to evaluate speech comprehension
skills in noisy environments, which individuals with hearing loss frequently encounter
in daily life, for two different signal processing techniques.
Method
Our study was performed in the Audiology, Language and Speech Disorders Clinic of
Istanbul University-Cerrahpaşa Faculty of Medicine, with the approval of the ethics
committee dated 08/10/2020 and numbered 132965. All participants were informed about
the methods to be applied voluntarily and their written consents were obtained.
Participants
Having at least 6 months of regular hearing aid use experience and in the 18 to 70
age range, 34 individuals with bilateral symmetrical mild to moderate-to-severe sensorineural
hearing loss (SNHL) were included in the study. The study involved 13 female and 21
male participants, with an average age of 51.47. Subjects with normal otoscopic and
immitansmetric findings and speech discrimination scores (SDSs) of 56% and above were
included in the study. All participants are multichannel hearing aid users for at
least 6 months (multichannel hearing aids dominate the sector and have the strongest
market penetration, so it was not possible for us to reach enough channel-free hearing
aid users).
Procedure
Within the scope of audiological evaluation, immitansmetric examination, pure-tone
audiometry, and speech tests were performed. After the evaluations, the individuals
who were determined to be suitable for the working conditions were included in the
hearing aid trial. Real ear measurements were performed and the fittings of the hearing
aids to be used were made. Free-field hearing assessment, Turkish matrix test (TMT),
spectral-temporally modulated ripple test (SMRT), and random gap detection test (RGDT)
were performed with multichannel and channel-free hearing aids after the necessary
hearing aid adjustments were made. After the fitting procedure of both hearing aids
was applied in random order, the performances were evaluated separately for all participants.
Audiological Assessment
Acoustic immitansmetric evaluations were performed with the GSI TympStar V.2 Middle-Ear
Analyzer (Grason-stadler Inc. Tiger/USA) device to evaluate middle ear functions of
all individuals participating in the study.
The tests applied to the participants for pure-tone audiometry and other evaluation
methods were performed in a soundproof room in accordance with the standards, using
the Otometrics-MADSEN Astera[2] (Natus Medical Inc., Taastrup, Denmark) computer-controlled multichannel clinical
audiometer. Pure-tone averages were obtained by averaging the hearing thresholds of
all participants in the range of 500 to 4,000 Hz, and the degrees of hearing loss
were determined according to the Clark's classification.[8] In speech audiometry evaluation, speech reception threshold (SRT), most comfortable
loudness (MCL), SDS, and uncomfortable loudness (UCL) tests were applied, respectively.
Hearing Aid Selection and Fitting
In our study, Oticon Opn 1 miniRITE-64 channels (Smorum, Denmark) as multichannel hearing aid and Bernafon Zerena 9 miniRITE (Bern, Switzerland) as channel-free hearing aid were used bilaterally. These
hearing aids were selected because they have the same characteristics according to
the technical specifications and factory output measurements reported by the manufacturer.
Oticon and Bernafon are hearing aid companies that are part of Demant Holdings. These
hearing aid manufacturers belonging to the same holding use similar technologies,
hardware architecture, and firmware. During the measurements, to ensure equality in
terms of hearing aid features, applications were performed by adjusting all features
in the same way, including noise reduction mode off, fixed/full directional microphone
mode, and NAL-NL2 gain algorithm in both hearing aids. Additional features such as
frequency lowering, wind noise reduction, etc., were disabled. The real ear measurement
applications of the participants were performed with the Aurical Free Fit (GN Otometrics
A/S, Denmark) device. Based on real ear responses, appropriate gain targets were set
for each patient, and hearing aids were programmed accordingly.
Free-Field Hearing Assessment with Hearing Aids
In free-field conditions, a warble tone stimulus was sent from a speaker at 0-degree
azimuth and 1-m distance, and hearing threshold (in the range of 500–4,000 Hz), SRT,
MCL, SDS, and UCL assessments were made with hearing aids.
Spectral-Temporally Modulated Ripple Test
SMRT Version 1.1.3 (www.ear-lab.org) software was used to evaluate the spectral resolution. The SMRT software was installed
on the laptop and the computer was connected to the clinical audiometer and the signals
were sent as calibrated. The test was conducted in a soundproof room via free-field
speakers. The stimuli were presented at 65 dB (A) with the listener 1 m away from
the speaker and the loudspeaker angle 0 degree.[9] Participants were seated in front of the computer screen. During the test, three
boxes numbered 1, 2, and 3 appear on the screen and the relevant box lights up red
when the stimulus is given. Participants were presented with three stimuli consisting
of two references and one target stimulus, and they were asked to choose which of
the three stimuli on the screen was perceived differently by pressing the relevant
number on the keyboard. Scores in test results are reported numerically by the software
as ripples per octave.
Random Gap Detection Test
RGDT was used to evaluate participants' temporal resolution skills. In the application
of the test, in subtest 1, while the stimulus was given in which the intervals were
presented gradually, from small to large, for the participants to practice; in subtest
2, stimuli were presented at random intervals at frequencies of 500, 1,000, 2,000,
and 4,000 Hz, and to the standard test part was passed. Audio files of the RGDT stimuli
were uploaded to the computerized audiometer and given to the participants with a
free-field speaker at 1 m distance and 0 degree azimuth. Stimuli were presented at
a most comfortable listening level (50 dB sensation level).[10] Participants were asked to indicate that they heard one or two voices, and each
response was marked on the RGDT form. In this way, it is aimed to determine the shortest
time interval that can be detected by the individual, namely the temporal acuity threshold.
The smallest gap (ms) determined for each of the 500 to 4,000 Hz frequencies tested
was detected. The participants' composite RGDT threshold for both hearing aids was
determined as the average of reported results across four test frequencies of 500
to 4,000 Hz.
Turkish Matrix Test
Adaptive noise and nonadaptive noise procedures of the TMT were used to evaluate speech
understanding skills in noise. Tests with both hearing aids were performed in the
free field with a speaker at a distance of 1 m and an azimuth angle of 0 degree. Both
noise and speech signals were presented in such a way that they come from in front
of the participant (the most difficult listening condition). TMT was administered
via the Oldenburg Measurement Application software and an audiometer that supports
this software.
In the adaptive procedure, the SRT in which 50% of the stimuli are correctly repeated
in noise is determined as the signal-to-noise ratio (SNR). The noise level for this
test was presented as a constant at the level of 65 dB. For each procedure, 20 sentences
of 5 words were presented and the participants were asked to repeat all the words
they understood.[11] Correctly repeated words were marked on the software by the researcher. The speech
level was automatically changed by the software until it finds the critical SNR threshold
(the value reached to distinguish 50%). The level of speech decreased as the participants
repeated the sentences correctly, while it increased automatically when they gave
incorrect answers. After all the sentences of the test are completed, the test result
is presented as SNR (dB) by the software. In the nonadaptive procedure, SI in certain
fixed SNRs is determined as the percent correct score. In our study, 0 dB SNR (challenging
listening condition) was used. The speech stimulus was sent by adding 40 dB to the
pure-tone averages and the noise level was adjusted on the computer according to the
determined SNR.
Statistical Analysis
According to the results obtained by applying normality tests to the data, appropriate
parametric (independent samples t-test, one-way analysis of variance) or nonparametric (Mann–Whitney U, Kruskal–Wallis) tests were selected and used, and their results were interpreted
statistically. Spearman's correlation analysis was applied to define the direction
and degree of relations between variables. All research data were analyzed with the
IBM SPSS Statistics 26.0 program at a p = 0.05 significance level. According to the GPower analysis performed, 96.9% power
was obtained for the research with n = 34 people at a significance level of 0.05.
Results
Right/left ear average pure-tone threshold values according to frequencies, air/bone
conduction pure-tone averages, and speech audiometry results of the participants are
shown in [Fig. 1]. No statistically significant difference was observed between the right and left
ears in pure-tone and speech audiometry results (p > 0.05).
Fig. 1 Pure-tone and speech audiometry findings of the participants. (*p < 0.05. Hz: hertz, dB: decibel, PTA: pure-tone average, SRT: speech reception threshold,
SDS: speech discrimination score, Mann–Whitney U test).
There was no statistically significant difference between the two hearing aids in
the results of free-field hearing assessment and speech audiometry according to the
frequencies applied with multichannel and channel-free hearing aids (p > 0.05) ([Fig. 2]).
Fig. 2 Evaluation of participants' free-field hearing with hearing aids. (*p < 0.05. Hz: hertz, dB: decibel, SRT: speech reception threshold, SDS: speech discrimination
score, Mann–Whitney U test).
No statistically significant difference was observed between the two hearing aids
in the SMRT results applied with multichannel and channel-free hearing aids (p > 0.05) ([Fig. 3]).
Fig. 3 Comparison of SMRT results applied with multichannel and channel-free hearing aids.
(*p < 0.05. RPO: ripples per octave, SMRT: spectral-temporally modulated ripple test,
NS: not significantly different, independent samples t-test).
In the RGDT results applied with multichannel and channel-free hearing aids, statistically
significant differences were observed between the two hearing aids for the 1,000 Hz
(p = 0.045), 2,000 Hz (p = 0.046), and composite RGDT (p = 0.001) thresholds. For the 1,000 Hz, 2,000 Hz, and composite RGDT thresholds, lower
RGDT thresholds were obtained with channel-free hearing aids compared with multichannel
hearing aids. There was no statistically significant difference between the two hearing
aids for the 500 and 4,000 Hz RGDT thresholds (p > 0.05) ([Fig. 4]).
Fig. 4 Comparison of RGDT results applied with multichannel and channel-free hearing aids.
(*p < 0.05, **p < 0.01, ms: millisecond, Hz: hertz, RGDT: random gap detection test, NS: not significantly
different, Mann–Whitney U test).
In the comparison of composite RGDT thresholds obtained with multichannel and channel-free
hearing aids according to the degrees of hearing loss, there was no statistically
significant difference between the two hearing aids in the mild degree (p > 0.05). Although there was no significant difference, lower composite RGDT thresholds
were observed with channel-free hearing aids compared with multichannel hearing aids.
Statistically significantly lower composite RGDT thresholds were observed with channel-free
hearing aids compared with multichannel hearing aids in groups with moderate (p = 0.037) and moderately severe (p = 0.046) hearing loss ([Table 1]).
Table 1
Comparison of composite RGDT results for multichannel and channel-free hearing aids
according to different parameters
|
Composite RGDT thresholds
|
Mean ± SD
(ms)
|
Min
|
Max
|
p
|
|
Mild hearing loss (n = 8)
|
Multichannel hearing aid
|
18.7 ± 8.2
|
5
|
40
|
0.092
|
|
Channel-free hearing aid
|
15 ± 10.2
|
2
|
40
|
|
Moderate hearing loss (n = 14)
|
Multichannel hearing aid
|
19 ± 10.9
|
2
|
40
|
0.037[a]
|
|
Channel-free hearing aid
|
14.5 ± 9.8
|
2
|
40
|
|
Moderately severe hearing loss (n = 12)
|
Multichannel hearing aid
|
18.1 ± 8.3
|
2
|
40
|
0.046[a]
|
|
Channel-free hearing aid
|
15.2 ± 8.8
|
2
|
40
|
|
55 years and younger
(n = 17)
|
Multichannel hearing aid
|
17 ± 7.8
|
2
|
30
|
0.025[a]
|
|
Channel-free hearing aid
|
13.9 ± 8.1
|
2
|
30
|
|
Above 55 years old
(n = 17)
|
Multichannel hearing aid
|
20.2 ± 10.6
|
2
|
40
|
0.012[a]
|
|
Channel-free hearing aid
|
15.9 ± 10.7
|
2
|
40
|
|
Female
(n = 13)
|
Multichannel hearing aid
|
20.5 ± 8.6
|
5
|
40
|
0.080
|
|
Channel-free hearing aid
|
17.3 ± 10.5
|
2
|
40
|
|
Male
(n = 21)
|
Multichannel hearing aid
|
17.4 ± 9.7
|
2
|
40
|
0.005[a]
|
|
Channel-free hearing aid
|
13.4 ± 8.6
|
2
|
40
|
|
1 year or less experience (n = 17)
|
Multichannel hearing aid
|
20.7 ± 10.9
|
2
|
40
|
0.326
|
|
Channel-free hearing aid
|
18.5 ± 10.3
|
2
|
40
|
|
More than 1-year experience (n = 17)
|
Multichannel hearing aid
|
16.5 ± 7.1
|
2
|
30
|
0.000[a]
|
|
Channel-free hearing aid
|
11.3 ± 7
|
2
|
30
|
Abbreviations: Max, maximum; Min, minimum; ms, millisecond; RGDT, random gap detection
test; SD, standard deviation.
Note: Mann–Whitney U test.
a
p < 0.05.
In the comparison of composite RGDT thresholds obtained with multichannel and channel-free
hearing aids according to the age of the participants, statistically significantly
lower composite RGDT thresholds were observed with channel-free hearing aids compared
with multichannel hearing aids in both individuals aged 55 years and younger (p = 0.025) and above 55 years of age (p = 0.012) ([Table 1]).
In the comparison of composite RGDT thresholds obtained with multichannel and channel-free
hearing aids according to the gender of the participants, while there was no statistically
significant difference between the two hearing aids in female participants (p > 0.05), lower composite RGDT thresholds were observed with channel-free hearing
aids compared with multichannel hearing aids. In male participants, statistically
significantly lower composite RGDT thresholds were observed with channel-free hearing
aids compared with multichannel hearing aids (p = 0.005) ([Table 1]).
In the comparison of the composite RGDT thresholds obtained with multichannel and
channel-free hearing aids according to the hearing aid use experience of the participants,
while there was no statistically significant difference between the two hearing aids
in participants with 1 year or less experience (p > 0.05), statistically significantly lower composite RGDT thresholds were observed
with channel-free hearing aids compared with multichannel hearing aids in participants
with more than 1-year experience (p < 0.001) ([Table 1]).
No statistically significant difference was observed between the two hearing aids
in the results of the SRT in noise and SI in noise within the scope of the TMT applied
with multichannel and channel-free hearing aids (p > 0.05) ([Table 2]).
Table 2
Comparison of Turkish matrix test results applied with multichannel and channel-free
hearing aids
|
Turkish matrix test
|
Multichannel hearing aid
|
Channel-free hearing aid
|
p
|
|
Mean ± SD
|
Min
|
Max
|
Mean ± SD
|
Min
|
Max
|
|
SRT in noise (SNR-dB)
|
–0.46 ± 2.49
|
–5.1
|
6.1
|
–0.39 ± 2.52
|
–4.9
|
4.4
|
0.912[a]
|
|
SI in noise (%)
|
65.91 ± 13.45
|
30
|
91
|
61.65 ± 13.63
|
32
|
88
|
0.199[b]
|
Abbreviations: Max, maximum; Min, minimum; ms, millisecond; SD, standard deviation;
; SI, speech intelligibility; SNR, signal-to-noise ratio; SRT, speech reception threshold.
a Mann–Whitney U test.
b Independent samples t-test.
Discussion
In our study, multichannel and channel-free hearing aids were compared in terms of
spectral resolution using SMRT, and an innovative approach was presented to the literature
in this respect. According to the results we obtained, no statistically significant
difference was observed between the two hearing aids in terms of SMRT scores (p > 0.05) ([Fig. 3]). In a study in which acoustic analyses of hearing aid signal outputs were performed,
channel-free hearing aids also showed spectral distortion, similar to multichannel
hearing aids. These results support our study findings.[7] Plomp has suggested that multichannel hearing aids assign different compression
ratios to different channels depending on hearing loss at each frequency, which reduces
spectral contrast and changes the spectral shape of speech, resulting in lower speech
recognition scores.[12] Schaub suggested that spectral contrast is preserved in channel-free hearing aids
as the gain is quickly adjusted according to the incoming signal.[13] Despite the different opinions in the literature, both hearing aids showed similar
performance in terms of spectral resolution in our study.
Kodiyath et al showed that individuals with hearing loss derive significant benefits
from channel-free hearing aids over multichannel hearing aids in providing temporal
cues for signal detection in background noise.[14] In a study comparing the temporal processing skills of 21 individuals with SNHL
with channel-free and multichannel hearing aids, except for comodulation masking release-comodulated
condition (CMR-CM), there was no statistically significant difference between the
two hearing aids for temporal modulation transfer function, gap detection test, and
CMR-uncomodulated condition. For the CMR-CM task, the channel-free hearing aid performed
statistically significantly better than the multichannel hearing aid. Moreover, in
the acoustic analysis of the hearing aid signal outputs, channel-free hearing aids
also showed temporal distortion, similar to multichannel hearing aids.[7] In our results, 1,000 Hz (p = 0.045), 2,000 Hz (p = 0.046), and composite RGDT (p = 0.001) thresholds were found to be statistically significantly lower with channel-free
hearing aids. This has shown that temporal resolution performance is better than multichannel
hearing aids. Although the 500 and 4,000 Hz RGDT threshold averages were also found
to be lower for channel-free hearing aids, they did not reach statistical significance
(p = 0.152 at 500 Hz and p = 0.293 at 4,000 Hz). Although differences were found in significance according to
frequencies, strong statistical significance with the lower threshold in composite
RGDT results clearly showed that the temporal resolution performance of channel-free
hearing aids was superior (p = 0.001) ([Fig. 4]). The reason for this is thought to be that channel-free hearing aids can make faster
gain adjustments according to frequency.
Plyler et al have revealed that listeners who prefer channel-free processing have
more hearing loss than listeners who prefer multichannel processing.[5] In comparison of the composite RGDT results of our study, while there was no statistically
significant difference between the two hearing aids in the mild hearing loss group
(p > 0.05), better temporal resolution performance was observed with channel-free hearing
aids, with statistically significantly lower composite RGDT thresholds, compared with
multichannel hearing aids in groups with moderate (p = 0.037) and moderately severe (p = 0.046) hearing loss ([Table 1]). When using multichannel hearing aids, listeners with more hearing loss will need
higher compression ratios to ensure the audibility of low-level signals and the comfort
of high-level signals. High compression ratios have been shown to increase temporal
envelope distortion and reduce spectral contrast.[15] As a result, it was thought that listeners with more hearing loss may have performed
better with channel-free hearing aids due to the additional temporal and spectral
distortion in the multichannel hearing aid.
Better temporal resolution performance was observed with statistically significantly
lower composite RGDT thresholds in channel-free hearing aids compared with multichannel
hearing aids for both young adults (p = 0.025) and older adults (p = 0.012). This has shown that age does not have a significant effect on the comparison
of different hearing aids. In addition, although there was no significant performance
difference between the two hearing aids in female participants (p = 0.080), the observation of better temporal resolution performance with channel-free
hearing aids in males (p = 0.005) has shown that gender may be related to signal processing technique ([Table 1]).
In the study of Kodiyath et al, channel-free hearing aids were preferred more by first-time
users than experienced users.[14] Mohan and Rajashekhar have shown that while first-time hearing aid users prefer
channel-free processing, experienced users prefer multichannel processing. However,
no performance difference was observed for first-time and experienced users.[7] In the comparison of the composite RGDT results we obtained, there was no statistically
significant difference between the two hearing aids in the participants with 1 year
or less of experience (p > 0.05); in participants with more than 1-year experience, better temporal resolution
performance was observed with statistically significantly lower composite RGDT thresholds
in channel-free hearing aids compared with multichannel hearing aids (p < 0.001) ([Table 1]).
The results of a study that applied Pascoe's high-frequency word list in noise and
hearing in noise test showed that scores were not significantly different between
multichannel and channel-free hearing aids.[5] In the results of another study on speech recognition in noise, it was determined
that channel-free hearing aid outperformed the 8-channel hearing aid at 0 dB SNR;
no difference was found at 10 dB SNR. No statistically significant difference was
observed in the study.[16] Mohan and Rajashekhar found that the performance between multichannel and channel-free
hearing aids did not show a statistically significant difference for speech perception
in noise test.[7] According to the results we obtained, no statistically significant difference was
observed between multichannel and channel-free hearing aids for SRT in noise and SI
in noise within the scope of TMT (p > 0.05) ([Table 2]). The results of our study are generally compatible with the literature. Unlike
the literature, mean scores for SI in noise was higher with multichannel hearing aids
than channel-free hearing aids but did not reach statistical significance. Although
multichannel hearing aids use syllable compression and channel-free hearing aids use
phonemic compression with gain adjustment up to approximately 20,000 times per second,
the use of channel-free processing did not significantly improve or reduce speech
perception ability compared with multichannel processing.
Conclusion
As a result, no significant difference was found between multichannel and channel-free
hearing aids for spectral resolution and speech understanding in noise (p > 0.05). In temporal resolution measurement, statistically significantly better performances
were observed with channel-free hearing aids (p < 0.05). It is thought that faster processing of the incoming signal in channel-free
hearing aids improves temporal resolution performance.
Performing the measurements with a larger number of participants for the subgroups
of hearing loss degree, age, gender, and hearing aid use experience and applying the
tests in different loudspeaker positions may allow the results to be evaluated in
a more detailed and consistent manner. It is anticipated that our study findings may
help hearing care professionals to choose the optimal hearing aid signal processing
technique according to different situations.
