Download Effects of Fast, Slow, and Adaptive Amplitude Compression on

J Am Acad Audiol 25:834–847 (2014)
Effects of Fast, Slow, and Adaptive Amplitude
Compression on Children’s and Adults’ Perception of
Meaningful Acoustic Information
DOI: 10.3766/jaaa.25.9.6
Andrea L. Pittman*
Ashley J. Pederson*
Madalyn A. Rash*
Abstract
Background: Fast- and slow-acting amplitude compression parameters have complementary
strengths and weaknesses that limit the full benefit of this feature to hearing aid users. Adaptive
time constants have been suggested in the literature as a means of optimizing the benefits of amplitude
compression.
Purpose: The purpose of this study was to compare the effects of three amplitude compression release
times (slow, fast, and adaptive) on children’s and adults’ accuracy for categorizing speech and environmental sounds.
Research Design: Participants were asked to categorize speech or environmental sounds embedded in
short trials containing low-level playground noise. Stimulus trials included either a high-level environmental sound followed by a lower-level speech stimulus (word) or a high-level speech stimulus followed by a
lower-level environmental sound. The listeners responded to the second (low-level) stimulus in each trial.
The two stimuli overlapped temporally in half of the trials but not in the other half. The stimulus trials were
processed to simulate amplitude compression having fast (40 msec), slow (800 msec), or adaptive
release times. The adaptive-compression parameters operated in a slow fashion until a sudden
increase/decrease in level required a rapid change in gain.
Study Sample: Participants were 15 children and 26 adults with hearing loss (HL) as well as 20 children
and 21 adults with normal hearing (NH).
Data Collection and Analyses: Performance (in % correct) was arcsine transformed and subjected to
repeated-measures analysis of variance with pairwise comparisons of significant main effects using Bonferroni adjustments for multiple comparisons.
Results: Overall, the performance of listeners with HL was poorer than that of the listeners with NH and
performance for the environmental sounds was poorer than for the speech stimuli, particularly for the
adults and children with HL. Significant effects of age group, stimulus overlap, and compression speed
were observed for the listeners with NH, whereas effects of stimulus overlap and compression speed
were found for the listeners with HL. Whereas listeners with NH achieved optimal performance with
slow-acting compression, the listeners with HL achieved optimal performance with adaptive compression.
Conclusions: Although slow, fast, and adaptive compression affects the acoustic signal in a subtle fashion, amplitude compression significantly affects perception of speech and environmental sounds.
Key Words: Amplification, amplitude compression, children, adults
*Arizona State University, Tempe, AZ
Andrea Pittman, Department of Speech and Hearing Science, Arizona State University, P.O. Box 0102, Tempe, AZ 85287-0102; E-mail: andrea.
[email protected]
This work has not been published in any form.
This project was funded by a grant from Oticon A/S.
834
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Fast, Slow, and Adaptive Amplitude Compression/Pittman et al
Abbreviations: AI 5 articulation index; ANOVA 5 analysis of variance; DSL 5 Desired Sensation Level;
HL 5 hearing loss; NH 5 normal hearing; WDRC 5 wide dynamic range compression
INTRODUCTION
L
ike many advances in hearing aid technology
during the last 25 yr, only a few of the benefits
of amplitude compression were apparent when
it was first introduced. Chief among the initial benefits
was the reduced need for the hearing aid user to adjust the
volume control as the listening environment changed.
When it was also shown that hearing aid users preferred
amplitude compression versus linear amplification, the
feature became standard in most commercially available
devices (Souza, 2002; Walden et al, 2000; Kam and Wong,
1999). The benefits of amplitude compression were later
determined to include improved tolerance for loud sounds
as well as a higher rate of successful adjustment to hearing aid use (Larson et al, 2000; Johnson et al, 2010).
The most popular application of this feature is wide
dynamic range compression (WDRC). In simple terms,
WDRC is a specific implementation of the dynamic and
static properties of amplitude compression in which
more gain is applied to low-level sounds that fall at
or below the listener’s hearing thresholds whereas less
gain is applied to higher-level sounds that may exceed
the listener’s comfort. To do this, WDRC involves a
number of static and dynamic parameters such as (1)
the input level at which gain is no longer applied linearly (i.e., compression kneepoint), (2) the change in
input level relative to the change in output level (i.e.,
compression ratio), and (3) the speed with which the
gain parameters are applied (i.e., attack/release time).
The first two parameters (compression kneepoint and
ratio) have received a fair amount of attention such that
their effects are fairly well understood (Souza, 2002;
Souza and Turner, 1999; Davies-Venn et al, 2009; Souza
et al, 2005). The impact of the dynamic parameters
(attack and release time) have received somewhat less
attention, although evidence suggests that they may significantly affect successful communication for some hearing
aid users (Stone and Moore, 2003; Lunner and SundewallThorén, 2007; Jenstad and Souza, 2005; Gatehouse et al,
2006).
By convention, the dynamic parameters of amplitude
compression are subdivided into fast- and slow-acting
categories denoting the time it takes for the compression parameters to reach their nominal gain values
for any level of input. Most commercially available devices have a fast attack time (,50 msec), although their
release times can differ substantially. Release times in
fast-acting compressors are generally less than 100
msec, whereas the release times of slow-acting compressors can be between 500 and 2000 msec or more. The
goal of fast-acting compression is to reduce the dynamic
range of continuous speech over the short term by operating fast enough to apply gain to low-level phonemes
(e.g., voiceless fricatives) and less gain to high-level
phonemes (e.g., vowels). The goal of a slow-acting system, on the other hand, is to limit the effects of compression on the short-term variations of speech and instead
modulate levels of vocal effort and environmental
sounds over the long term.
Both of these approaches cause unique problems for
the hearing aid user. Slow-acting compression with its
longer release times can result in inadequate amplification of low-level input for a substantial period following
a loud input. Fast-acting compression can cause two problems: (1) it can distort the acoustic waveform resulting in
a somewhat poorer signal quality (Jenstad and Souza,
2005), and (2) it can decrease the signal-to-noise ratio
by amplifying low-level background noise during periods
of low-level speech (Naylor and Johannesson, 2009).
Recent evidence comparing the effects of fast- and slowrelease times on speech perception suggests that adults
with lower cognitive abilities have more difficulty with
the subtle signal distortion caused by fast-acting compression than adults with better cognitive abilities (Gatehouse
et al, 2006; Souza et al, 2000; Jenstad and Souza, 2005,
2007; Cox and Xu, 2010; Lunner and Sundewall-Thorén,
2007; Rudner et al, 2011). To date, no studies examining
the effects of compression speed have been conducted in
children with hearing loss (HL) who may be vulnerable to
the distortion caused by fast-acting compression while
their cognitive skills are developing.
Another area that has received little attention is the
relationship between the acoustic event that triggers
the compression and the acoustic event that the hearing
aid user would like to hear. Most research has focused
on the effects of compression speed on the speech signal.
However, complex listening environments in which
amplitude compression would likely occur contain multiple sound sources and levels of acoustic information. For
example, walking along a city street may require the
hearing aid user to manage environmental sounds
(e.g., traffic noise), multitalker babble from other pedestrians, and an interaction with a conversation partner.
Any of these sound sources could be of interest to the
hearing aid user at any point in time, although the loudest sound source (usually closest) to the hearing aid is
likely to trigger compression, even speech. Thus, the
manner in which the hearing aid applies compression
to both speech and environmental sounds may be quite
important to the hearing aid user.
Several things have contributed to the limited
amount of research in the perception of environmental
sounds:
835
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Journal of the American Academy of Audiology/Volume 25, Number 9, 2014
(1) It is more challenging to represent the wider variety
of environmental sounds in an experimental paradigm than it is to represent the more narrow range
of speech sounds (Shafiro and Gygi, 2004).
(2) Individuals with severe-to-profound HLs are more
likely than individuals with lesser HL to express
concern about their lack of access to environmental
sounds (Tyler, 1994; Zhao et al, 1997; Looi and
Arnephy, 2010).
(3) Environmental sounds are often confused with undesirable noise levels associated with public health concerns (Kim et al, 2012; Smith, 1998; Shepherd et al,
2010), causing them to be viewed as dangerous
rather than informative.
(4) Humans do not interact with environmental sounds
in the same way that they interact with speech
(Gygi and Shafiro, 2007). When necessary, listeners
respond to environmental sounds by modifying
their behavior rather than engaging with the sound
source as they might with another individual. Thus,
a lack of awareness for environmental sounds may
go unnoticed until an individual fails to modify his
or her behavior appropriately in an important situation (e.g., not responding to an alerting device, being
startled by an approaching person or vehicle).
Environmental sound recognition has been studied in
related disciplines to evaluate such things as awareness
of, and tolerance for, sounds in the environment (Dawson
et al, 2004; Thawin et al, 2006; Dockrell and Shield,
2004; Spaulding et al, 2008); pattern recognition in
autistic children (van Lancker et al, 1988); and the
development of left- versus right-ear dominance in children (Kraft et al, 1995; Kraft, 1982). Fabiani et al (1996)
used an environmental-sounds naming paradigm to
examine, among other things, the effects of developing and aging cognition. Their experimental groups
included typically developing children (ages 5–7, 9–
11, and 14–16 yr) as well as younger (ages 19–34 yr)
and older adults (ages 61–88 yr). Using somewhat strict
naming criteria for the environmental sounds, they
found a significant increase in naming accuracy as a
function of age that reached adult-like performance
by ages 14–16 yr. Performance decreased significantly
for the older adults such that their naming accuracy
was equivalent to that of the youngest children.
Although the reasons for the poorer performance by
the youngest and oldest listeners were not addressed
directly, the children may have been less familiar with
environmental sounds because of inexperience, whereas the mild to moderate HLs reported for the older
adults may have reduced their ability to recognize the
sounds.
One study examined the effect of hearing aid signal
processing (frequency lowering) on listeners’ recognition of environmental sounds (Ranjbar et al, 2008); how-
ever, the participants were children and adults with
normal hearing (NH), so the results cannot be applied
with confidence to individuals with HL. There are currently no data regarding the effects of amplification on
the identification of environmental sounds in children
or adults with mild to moderately severe HL. To determine the impact of HL on environmental sound identification, the same listening conditions would need to be
applied to both NH and hearing-impaired listeners.
Referring back to the auditory scene of the busy city
street described earlier, a hearing aid user may need
to respond to any number of sounds in the environment
either indirectly (i.e., modifying his or her behavior) or
directly (i.e., interacting with the sound source). In this
environment, both fast- or slow-acting compression
may cause problems for the hearing aid user. Specifically, fast-acting compression will effectively reduce
the startling effects of sudden and loud sounds (e.g.,
shout, car horn) but will amplify lower-level street noise
such that the effective signal-to-noise ratio will be
poorer than the nominal signal-to-noise ratio. Put
another way, the listening environment will sound noisier to a hearing aid user than to someone without a
hearing aid (Naylor and Johannesson, 2009). On the
other hand, slow-acting compression will respond well
to sudden high-level sounds, but the longer release time
will cause the amplitude of later sounds to be reduced
such that the hearing aid user will receive less-thanadequate audibility for a period (i.e., the duration of
the release time). To address these complex situations,
Moore et al (2010) described a third option in the form of
an amplitude compressor with adaptive-time constants.
Such a system is “… usually slow-acting but the gain is
rapidly reduced by the fast system in response to sudden large increases in sound level. If the increase in
sound level lasts for only a short time, then the gain rapidly returns to the value set by the slow-acting system.
This provides protection from brief intense transients,
with little effect on the long-term gain.” (p. 369). With a
device using both slow- and fast-acting compression, the
hearing aid user may have a better chance of recognizing lower-level acoustic events (speech or environmental sounds) that occur in close temporal proximity to
loud acoustic events.
In the present study, the effects of three configurations of amplitude compression (slow-, fast-, and adaptivetime constants) were examined in regard to the
perception of meaningful acoustic stimuli in complex
listening environments. The purpose of the study was
to determine the manner in which the three forms of
compression affect children’s and adults’ accuracy for
categorizing speech and environmental sounds, particularly when the stimulus is partially masked by
another stimulus responsible for triggering the amplitude compression. It was hypothesized that the performance of the adults would exceed that of the
836
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Fast, Slow, and Adaptive Amplitude Compression/Pittman et al
children and that the listeners with NH would perform
better than those with HL. It was also hypothesized
that performance with the adaptive-time constants
would exceed performance for the slow- or fast-time constants. These results were expected to be most apparent
for partially masked stimuli in which only a few milliseconds of information is available.
METHOD
Participants
Participants were 21 adults between 21 and 30 yr old
(mean age 5 24 yr) with NH and 26 adults between 50
and 78 yr old (mean age 5 66 yr) with permanent HL
(Table 1). (Originally, the adults with HL were divided
into two age groups (age ranges 50–65 yr and 66–78 yr)
in an effort to detect age effects. The groups were combined when it was determined that performance and
age were not significantly related.) There were 20 children with NH and 15 children with permanent HL who
also participated. Both groups were between 7 and 12 yr
old (mean age 5 10 yr). At the time of testing, the outer
and middle-ear status of each participant was determined
using otoscopy and tympanometry, respectively, and confirmed to be consistent with the existing underlying aural pathologic features. Pure-tone thresholds were also
obtained using a standard clinical audiometer and insert
earphones. All testing was conducted in a sound-treated
booth meeting American National Standards Institute
(ANSI) standards for room noise (ANSI, 1999). The participants with NH had pure-tone thresholds of 20 dB
HL or less for octave frequencies between 250 and 8000
Hz. Figure 1 shows the average (61 SD) hearing thresholds for the right and left ears of the children and adults
with HL in the upper and lower panels, respectively. On
average, the participants with HL had mild to moderately
severe HLs of similar configurations. The results are
shown in units of dB SPL to facilitate the display of the
stimulus levels (dashed lines) in a format consistent with
that used during hearing aid fitting (discussed below).
Stimuli
Speech Stimuli
The speech stimuli were three lists of 36 common
words containing equal numbers of unique items referring to people, foods, and animals [see McFadden and
Pittman (2008) and Pittman (2011) for a detailed
description of these stimuli]. All of the words were
nouns and were within the vocabulary of first-grade
children (Moe et al, 1982). The words varied in length
from one to three syllables and were spoken by a female
talker having a standard American-English dialect.
Digitized recordings were made at a sampling rate of
Figure 1. Average right and left hearing thresholds using standard audiometric symbols are shown in units of sound pressure
level (dB SPL) as a function of frequency for the children with
HL (upper panel) and the adults with HL (lower panel). Error bars
represent 61 SD. Also shown is the average level of the stimuli
presented to the right and left ears of each listener (dashed lines).
22.05 kHz and edited to create individual, normalized
audio files (Adobe Audition, CS5.5). The appendix contains the speech stimuli sorted alphabetically by category. It should be noted that the relative difficulty of
the stimulus lists is not known.
Environmental Sound Stimuli
The environmental sounds also comprised three sets of
36 unique sounds representing biologic, mechanical, and
animal sound sources. The sounds were drawn from multiple sources including several websites (www.findsounds.
com, www.sound-ideas.com, www.hollywoodedge.com,
sound-effects-library.com), those available through a
royalty-free sound-effects feature in Adobe Audition
(CS5.5), and those generously shared by Dr. Valeriy
Shafiro in the Auditory Research Laboratory at Rush
University Medical Center, Chicago, IL. All files were
converted to wave file format having a single (mono)
channel with a sampling rate of 22.05 kHz and 32-bit
resolution and normalized. The appendix contains the
complete list of sounds in each category sorted alphabetically. As with the speech stimuli, the relative difficulty of the stimulus lists is not known.
837
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Journal of the American Academy of Audiology/Volume 25, Number 9, 2014
half (lower panel). The expected effects of these parameters on performance are described below.
Acoustic Competitor
Playground noise was used as an acoustic competitor
to the speech and environmental sounds. This competitor was obtained from Adobe Audition (CS5.5) and contained sounds consistent with a neighborhood park or
recreation area where children are playing.
Stimulus Trials
Each trial contained a speech stimulus, an environmental sound, and the playground noise as depicted
in the upper and lower panels of Figure 2. The upper
waveform (trigger) represents a high-level stimulus
designed to trigger the amplitude compression. The
middle waveform (target) represents the stimulus to
which the listener was instructed to respond. The lower
waveform represents a low-level playground noise that
was presented continuously throughout each 1.5–3 sec
trial. The target was presented at a conversational level
of 65 dB SPL, whereas the presentation level of the trigger and the playground noise was 6 dB above and 4 dB
below that of the target, respectively. Thus, the target,
triggering, and playground stimuli were presented
at 65, 71, and 61 dB SPL, respectively. Put another
way, the target and trigger stimuli were 10 and 4 dB
above the playground noise, respectively. Figure 2 also
shows the two temporal arrangements of the triggering
and target stimuli. In half of the trials, the two stimuli
were separated by z400 msec (upper panel), whereas in
the remaining trials the two stimuli overlapped by roughly
Figure 2. Schematic of the complex stimuli for the nonoverlapping and overlapping trials.
Amplitude Compression Speed
The stimuli were processed digitally by engineers at
Oticon A/S using conventional parameters for slowacting (attack time z20 msec, release time z800 msec)
and fast-acting (attack time z10 msec, release time
z40 msec) compression, as well as a custom adaptivecompression strategy specific to this manufacturer’s
devices (i.e., SpeechGuard). The adaptive compression
strategy uses two internal level estimators (one for
increases in input level and one for decreases in input
level) as well as a difference detector. The difference
detector tracks the magnitude of the level changes
and computes new time constants accordingly. When
there are no large changes in the input level, longer
time constants are applied (.800 msec). When a large
increase in input level occurs, short-time constants are
applied, which causes an instantaneous, short-term
reduction in gain. When a large decrease in input level
is detected, slow-time constants are again applied. This
allows the system to quickly follow increases and decreases
in input level. A nominal compression ratio of 3:1, a knee
point of 40 dB, and 4 channels having crossover frequencies at 625, 1562.5, and 3125 Hz were used in all
three compression schemes for an overall stimulus level
of 65 dB SPL. The nominal compression ratio was
limited to 3:1 to impose some, but not extensive, distortion to the amplitude variations between vowels and consonants (Hornsby and Ricketts, 2001; Boike and Souza,
2000), as would be expected in a fast-acting system. No
frequency shaping or maximum power output (OSPL90)
parameters were applied during digital processing.
Figure 3 shows the effects of the compression parameters on the environmental and speech stimuli. Each
panel shows the envelope of the original waveform
(dashed line) and the envelope of the compressed waveform (solid line). In this example, the stimuli were a
birdcall overlapping with the word “lifeguard.” The
amplitude of the second segment of the birdcall is twice
that of the first, and the lowest amplitude occurs during
the fricative “f ” in the word “lifeguard.” The solid line
represents the envelope of the waveform after the
slow-, fast-, and adaptive-compression parameters were
applied and are shown in the upper, middle, and lower
panels, respectively. All three compression parameters
effectively lowered the amplitude of the birdcall, but the
gain applied to the word “lifeguard” differed substantially. As expected, slow compression maintained the
amplitude variations within the waveform but decreased
the level overall. Fast compression amplified the playground noise and the target but reduced the amplitude
variations within the waveform and decreased the overall signal-to-noise ratio. Adaptive compression amplified
838
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Fast, Slow, and Adaptive Amplitude Compression/Pittman et al
Figure 4. Normalized amplitude envelopes of the fast (black),
slow (red), and adaptive (blue) compression stimuli.
Figure 3. Normalized amplitude envelopes of the original (dashed
line) and compressed stimuli (solid line) for the slow (upper panel), fast
(middle panel), and adaptive (lower panel) compression conditions.
the target and preserved the amplitude variations within
the waveform. Figure 4 shows the compressed stimuli displayed together in one panel such that the relative amplitude of each segment of the waveforms is revealed.
Referring back to Figure 2, the nonoverlapping and
overlapping arrangement of the stimuli were designed
to reveal the listener’s perception of partial and wholetarget stimuli with respect to the unique characteristics
of the three amplitude compression speeds. That is, the
same compression effects were expected for both configurations of the stimuli, but the listeners received only a
partial stimulus in the overlapping condition. Performance in the overlapping condition was expected to
reveal the compression condition that best preserved
the integrity of the target stimulus.
Stimulus Presentation
The stimuli were presented binaurally under earphones
(Sennheiser 25D) having a flat frequency response to
10 kHz. The presentation level of the target stimuli was
set to a conversational level (65 dB SPL) for the listeners
with NH. Custom laboratory software was used to shape
the long-term spectra of the stimuli to accommodate the
hearing thresholds of each listener with HL. Amplification targets for average conversational speech were calculated using the Desired Sensation Level (DSL) v5.0
fitting algorithm (Seewald et al, 1997). The age, transducer, and hearing thresholds of each participant were
entered into custom DSL v5.0 laboratory software provided by members of the Child Amplification Laboratory
at the University of Western Ontario. Figure 1 shows the
average frequency shaping (dashed lines) relative to the
hearing threshold of the children and adults in the upper
and lower panels, respectively. The hearing thresholds
were obtained under earphones (insert or supra-aural)
and converted to dB SPL relative to the appropriate coupler (2 cm3 or 6 cm3, respectively).
The audibility of the frequency-shaped stimuli was
calculated relative to the listener’s hearing thresholds
using the following formula, which is a modified version
of the speech intelligibility index (ANSI, 1997):
AI5
18
1 X
½Li hi W i
30 i51
ð1Þ
In this formula, i is the center frequency of each thirdoctave band between 0.1 and 8 kHz, u is the hearing
threshold in dB SPL (interpolation was applied at interoctave frequencies), and L is the level of the amplified
speech and environmental sounds (in dB SPL). These
values were taken from the long-term spectrum of the stimulus transduced by a supra-aural earphone (Sennheiser
25D) and measured in a 6 cm3 coupler. The long-term
spectrum was composed of 18 third-octave bands analyzed with a 40 msec Hanning window (50% overlap).
(The long-term spectra of the speech and environmental
sounds differed by ,3 dB at frequencies between 0.2
and 8 kHz.) Sensation levels (L-u) across the thirdoctave bands were restricted to a range of 0–30 dB
and multiplied by the importance of each band (W) using
weights for average speech (ANSI, 1997), and summed.
839
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Journal of the American Academy of Audiology/Volume 25, Number 9, 2014
Unfortunately, a 6 cm3 coupler-to-real-ear transform
does not exist, so the resulting articulation index (AI)
values are a close approximation to real-ear sound pressure levels.
AI values were calculated for each ear of each individual and averaged to provide one AI value per listener.
These values range from 0–1 with 0 indicating that the
stimuli were not audible (i.e., they were below threshold), whereas an AI value of 1 indicates that the stimuli
were fully audible. Because no data are available regarding the audibility required to optimize environmental
sounds, amplification parameters were set to maximize
the audibility of speech. Those values were referenced to
Stelmachowicz et al (2000), who examined the relationship between speech perception and audibility for both
children and adults with NH as well as children with
HL. Their results showed that by age 8 yr, perception
scores of at least 90% can be expected for AI values of
0.5. In the present study, the AI values calculated for
the children and adults with HL were 0.86 (SD 5
0.12) and 0.77 (SD 5 0.15), respectively. These AI values
were considered to be sufficient to optimize performance
such that the subtle acoustic effects of amplitude compression could be detected.
Procedures
Listeners participated in a total of 6 test conditions
(3 compression conditions 3 2 target stimuli) having
36 trials each. The experimental task was to correctly
identify as many (a) speech stimuli or (b) environmental
sound stimuli as possible under different test conditions. Performance (in % correct) was calculated for
each participant in each test condition. For three of
the test conditions, an environmental sound triggered
the compression, whereas in the other three listening
conditions, a speech stimulus triggered the compression. Half of the 36 trials in each condition contained
overlapping stimuli whereas the stimuli did not overlap
in the other half of the trials. The overlapping and nonoverlapping trials were randomly presented within
each listening condition so that the temporal pattern
of the trigger and target stimuli would not become
familiar to the listeners and aid their performance.
The listeners were instructed to ignore the triggering
stimulus (the first sound heard) and indicate the category to which the target stimulus belonged (the second
sound heard). A visual aid was provided to remind the
listeners of the response categories (i.e., person/food/
animal, person/animal/thing). No trials contained pairs
of waveforms from the same categories, although one or
two interesting combinations of words and environmental sounds slipped through the quality-control process
before data collection (horse neighing followed by the
word “cowboy”; rat squeaking followed by the word “lawyer”). Listeners were given the option to say, “I don’t
know” or to guess if they were unable to respond with
confidence. This effectively reduced the rate of chance
performance to 25% or less. The order of the six listening
conditions was counterbalanced across participants to
distribute learning effects equally across the conditions.
One of the challenges in adult-child research is finding a task for both children and adults that is not too
easy or too difficult for either group. The problem is further confounded when the groups include listeners with
normal and impaired hearing. To moderate the performance of each group, the listeners were also asked
to perform a visual task while responding to the auditory stimuli. At the start of each session, the listener
was handed an 8-1/2 3 11-inch page containing 20 rows
of patterns of simple shapes (3 of the patterns are shown
in Figure 5). The listener was instructed to solve the
pattern and circle the shape that would occur next from
the choices provided at the end of each row. Approximately two pages of patterns were completed per condition. The visual task served as a competitor to the
auditory stimuli and prevented performance (particularly for the listeners with NH) from reaching a ceiling
in all listening conditions. Also, the visual task occupied
the attention of the listeners such that reinforcement
for correct or incorrect responses to the auditory stimuli
was not expected or provided. Previous research has
shown that performance for similar visual tasks is
not affected by the difficulty of the auditory tasks in
children with NH and HL (Pittman, 2011). Therefore,
the visual task was used as a competitor to the auditory
task only and was not scored.
In accordance with the policies of the Internal Review
Board at Arizona State University, informed consent
was obtained from the adult listeners before testing.
Likewise, parental consent and child assent were obtained
for the children. Testing required no more than 2 hr, and
each listener was paid for his or her participation.
RESULTS
T
he data for each listening condition were first separated into overlapping and nonoverlapping trials
and were then reduced to a percent correct score. These
values were then arc-sine transformed to equalize the
variance across the range of scores (Studebaker, 1985)
and were averaged across groups and conditions. Analyses included an examination of the results for the listeners with NH to determine the effects of age and
Figure 5. Example of the visual pattern-completion task.
840
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Fast, Slow, and Adaptive Amplitude Compression/Pittman et al
stimulus condition. The overall performance of the listeners with NH and HL was then compared to identify
effects of HL. Finally, the performance of the listeners
with HL was examined in greater detail to determine
the effects of compression and overlap for each type
of stimulus. This approach provided a framework with
which to assess the performance of the listeners with
HL relative to the optimal performance of the listeners
with NH so that reasonable expectations may be made
regarding the degree to which performance was affected
by HL and stimulus condition.
HL versus NH
Figure 6 shows the results for the speech stimuli
in the upper panel and for the environmental sounds
in the lower panel for the listeners with NH. The
hatched-and-filled bars represent performance in the
nonoverlapping and overlapping conditions, respectively.
Each set of bars represents a different amplitude compression condition. As expected, performance was poorer
for the overlapping stimuli, and the children performed
more poorly than the adults. These observations were
examined via repeated-measures analysis of variance
(ANOVA) with age (adults, children) as the between-
Figure 6. Average performance of the listeners with NH for the
speech and environmental sound categorization task shown as a
function of age group. The parameter in each panel is compression
and overlapping stimulus condition.
participant factor and overlap (nonoverlapping, overlapping), compression speed (slow, fast, adaptive), and
target stimuli (speech, environmental sounds) as the
within-participant factors. Significant main effects of
age [F(1,39) 5 49.9, p , 0.001, h2 5 0.562, b 5 1], overlap
[F(1,39) 5 178, p , 0.001, h2 5 0.821, b 5 1], speed [F(1,78) 5
3.353, p 5 0.04, h2 5 0.79, b 5 0.617], and stimuli
[F(1,39) 5 101, p , 0.001, h2 5 0.723, b 5 1) were
revealed. Also, significant speed 3 overlap [F(2,78) 5
15.6, p , 0.001, h2 5 0.286, b 5 0.99] and stimuli 3
age [F(1,39) 5 5.9, p 5 0.02, h2 5 0.132, b 5 0.659] interactions were found. All other interactions were not significant. These results confirm that, for the listeners
with NH, performance was significantly affected by
the age of the listener (adults . children), the degree
to which the stimulus was masked by the trigger (nonoverlapping . overlapping), and the type of stimuli
(speech . environmental sounds). As for the significant
main effect of compression speed, pairwise comparisons
(with Bonferroni adjustment) revealed that performance was significantly higher for the slow compression
condition relative to fast and adaptive compression.
Figure 7 shows the results for the listeners with HL
using the same convention as in Figure 6 and shows that
overall, the performance of the listeners with HL was
poorer than their NH counterparts. Separate univariate
Figure 7. Average performance for the listeners with HL using
the same convention as in Figure 6.
841
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Journal of the American Academy of Audiology/Volume 25, Number 9, 2014
Table 1. Demographic Characteristics of the Children and Adults with HL including Gender, Age (in yr), and Hearing
Thresholds (in dB HL)
Right Ear
#
Left Ear
M/F
Age
0.25
0.5
1
2
4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
M
M
M
F
F
F
F
M
M
F
F
F
M
F
M
M
M
M
F
F
F
M
F
M
F
F
F
M
M
F
F
7
7
9
9
9
11
11
12
12
12
9
72
65
68
59
60
62
64
67
69
69
76
78
61
63
76
60
62
64
50
70
40
40
45
5
25
25
10
10
25
5
30
30
25
35
20
55
30
35
60
5
25
40
25
20
15
20
70
15
20
45
40
40
50
55
20
30
35
15
25
35
10
35
30
35
35
15
65
35
25
60
10
25
55
20
20
20
45
60
15
25
50
50
45
65
65
50
40
50
10
40
45
35
40
45
40
45
10
65
45
35
65
35
45
75
30
30
50
45
50
15
50
65
55
55
65
90
65
35
55
45
50
45
50
55
55
60
45
10
70
60
55
60
55
55
75
55
45
55
50
50
30
75
65
55
50
65
70
45
30
45
60
55
35
45
45
65
95
55
70
80
60
60
65
55
65
80
65
55
45
45
60
50
70
60
55
32
33
34
35
36
37
38
39
40
41
F
F
M
M
M
M
M
F
F
F
9
10
12
12
60
54
70
68
78
75
40
5
10
35
35
20
35
10
20
25
45
5
20
40
45
20
35
10
15
20
50
0
40
50
50
25
25
15
25
40
50
5
55
50
45
40
25
20
20
55
50
25
50
55
60
60
40
35
20
60
8
0.25
0.5
Hearing Device
1
2
4
8
40
35
40
45
60
40
50
60
55
45
55
70
10
10
20
50
30
25
35
30
45
25
45
50
55
10
5
10
70
10
25
40
50
25
35
45
45
5
10
35
35
30
35
50
55
20
20
45
NR
25
35
45
60
35
35
35
75
20
25
15
85
60
70
70
50
35
50
40
85
80
65
65
90
60
60
60
50
10
15
25
70
30
40
45
70
55
75
75
75
10
15
25
50
20
15
25
45
20
25
45
65
25
35
45
75
45
40
40
50
20
20
25
70
20
20
35
55
40
45
55
65
45
55
55
Not Using Amplification
45
35
40
50
45
5
0
5
70
10
15
25
40
30
35
50
45
35
45
50
85
20
20
25
50
25
25
25
45
10
10
15
65
20
15
15
65
20
25
35
55
65
70
65
30
50
45
80
40
40
55
65
50
50
25
65
75
55
60
50
50
80
35
55
55
45
30
35
55
75
50
50
60
80
45
25
65
55
70
40
45
50
65
75
55
75
80
65
80
60
55
80
75
55
55
50
55
60
40
60
70
60
40
65
65
10
25
65
60
75
50
45
40
55
90
60
70
80
60
NR
85
60
75
75
70
50
50
65
60
40
70
60
80
45
0
40
55
45
35
35
15
15
55
50
30
50
50
60
55
45
50
20
70
45
55
65
45
50
65
55
40
65
75
Make
Model
Style
O
P
P
P
P
P
P
P
P
S
W
CC
I
MT
O
O
O
O
O
O
P
P
P
RS
RS
RS
R
ST
ST
W
W
Safari P
Solana SP
Exelia Art P
Nios S H20 III
Cassia M H20
Solana Micro P
Nios Micro III
Nios H20 V
Solana Micro P
Destiny 1200
AK-9
KSHA03
IIQ43
Seneca
Agil Pro
Alta P
VPC-I
Acto Pro
unknown
Syncro V2
Audeo Smart V
Ambra Micro P
Versata M
Pulse
Forza
unknown
RX34
VQ820
VQ720
Mind 440
Senso Diva
BTE
BTE
BTE
BTE
BTE
BTE
RITA
ITE
BTE
BTE
BTE
RIC
ITE
BTE
RIC
BTE
ITE
RIC
BTE
BTE
RIC
BTE
BTE
RITA
RIC
RIC
ITE
RIC
RIC
ITE
BTE
Note: Also shown is the make, model, and style of hearing devices for those who used amplification. BTE 5 behind the ear; CC 5 Costco; I 5
Interon; ITE 5 in the ear; MT 5 MicroTech; NR 5 no response; O 5 Oticon; P 5 Phonak; R 5 Rexton; RIC 5 receiver in the canal; RITA 5 receiver
in the aid; RS 5 ReSound; S 5 Starkey; ST 5 SeboTek; W 5 Widex.
ANOVAs for the speech and environmental sounds were
performed to compare the overall performance of the NH
and HL groups. For each analyses, hearing group (NH,
HL), age (adults, children), and overlap (nonoverlapping, overlapping) were entered as fixed factors.
For the speech stimuli, significant main effects were
found for hearing group [F (1,484) 5 455, p , 0.001],
age [F(1,484) 5 16.9, p , 0.001], and overlap [F(1,484) 5
338, p , 0.001], as well as significant hearing group 3
age [F(1,484) 5 34.7, p , 0.001] and hearing group 3 overlap [F(1,484) 5 9.6, p 5 0.002] interactions. Not significant
were the age 3 overlap and age 3 hearing 3 overlap
interactions. For the environmental sounds, significant
main effects were found for hearing group [F(1,484) 5
436, p , 0.001] and overlap [F(1,484) 5 116, p , 0.001]
but not for age [F(1,484) 5 3.8, p 5 0.053]. A significant
842
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Fast, Slow, and Adaptive Amplitude Compression/Pittman et al
hearing group 3 age [F(1,484) 5 34.7, p , 0.001] interaction was found with all other interactions being
nonsignificant. These analyses revealed that (1) the
performance of the listeners with HL was significantly
poorer than that of the listeners with NH, (2) the children performed more poorly than the adults with the
speech stimuli but not the environmental sounds,
and (3) all groups performed more poorly in the overlapping condition. The group 3 age interaction for
both the speech stimuli and environmental sounds
indicates that the largest difference in performance
occurred between the adults with NH and the adults
with HL. Overall, these analyses indicate that nearly
every stimulus and age parameter served to reduce
the performance of the listeners with HL relative to
their NH counterparts.
HL versus Compression Speed
Four repeated-measures ANOVAs were conducted to
examine the interaction between HL and compression
speed for each stimulus and overlap condition. In each
analysis, age (adults, children) and compression speed
(slow, fast, and adaptive) were entered as between- and
within-participant factors, respectively. For significant
effects of compression speed, post hoc analyses were
conducted with Bonferroni adjustment for multiple
comparisons. Starting with the nonoverlapping speech
stimuli (hatched bars in the upper panel), a significant main effect for compression speed was observed
[F(2,78) 5 14.278, p , 0.001, h2 5 0.268, b 5 0.998] but
not for age [F(1,39) 5 7.14, p 5 0.403]. Nor was a speed 3
age interaction revealed [F(2,78) 5 1.722, p 5 0.185].
Pairwise comparisons revealed a significant difference
between the slow and fast compression speeds (I-J 5
0.162, p 5 0.012). These results indicate that performance was equally good in both the slow and adaptive
compression conditions when the stimuli were not
masked by the trigger and that the children with HL
performed as well as their adult counterparts. Similar
main effects and interactions were found for the overlapping condition (filled bars in the upper panel). A significant main effect of compression speed [F(2,78) 5
27.048, p , 0.001, h2 5 0.410, b 5 1] was found but
no effect of age [F(1,39) 5 0.006, p 5 0.940) or significant
age 3 compression speed [F(2,78) 5 1.937, p 5 0.151]
interaction. Pairwise comparisons of compression speed
revealed significantly better performance for adaptive
compression compared with both slow (I-J 5 0.193,
p , 0.001) and fast (I-J 5 0.199, p , 0.001) compression.
These results revealed significantly higher performance with the adaptive compression when the target
stimulus was partially masked by the triggering stimulus. Also, performance in the slow and fast compression conditions was equally poor for both the children
and adults.
Similar results were again observed for the environmental sounds in the nonoverlapping stimuli (hatched
bars in the lower panel). A significant main effect of
compression speed [F(2,78) 5 10.163, p , 0.001, h2 5
0.207, b 5 0.983] was revealed but not for age [F(1,39) 5
7.20, p 5 0.401]. For these stimuli, however, no speed 3
age interaction was revealed [F(2,78) 5 1.916, p 5
0.154]. Pairwise comparisons revealed significantly better performance for adaptive compression compared
with the slow (I-J 5 0.120, p 5 0.048) and fast (I-J 5
0.210, p , 0.001) compression speeds. Thus, performance was best in the adaptive compression condition
when the stimuli did not overlap. As for the overlapping
stimuli (filled bars in the lower panel), no main effects of
compression speed [F(2,78) 5 1.706, p 5 0.188] or age
[F(1,39) 5 0.354, p 5 0.555] were found. However, a significant age 3 compression speed interaction [F(2,78) 5
4.726, p 5 0.012, h2 5 0.108, b 5 0.775] was revealed.
Pairwise comparisons failed to reveal significant differences between the compression conditions, indicating that the interaction was likely the result of the
difference in performance between groups for the fast
compression condition. That is, the adults and children performed similarly in the slow and adaptive
compression conditions but not in the fast compression condition.
Relationship between Speech and
Environmental Sound Performance
It was of interest to determine the relationship between
the listener’s performance for speech and environmental
sound categorization. Shafiro (2008) and other authors
have argued that the acoustic and semantic properties
of speech and environmental sounds can be exploited to
further understand the difficulties that a listener with
HL might be experiencing with auditory perception
(Fabiani et al, 1996; Gygi and Shafiro, 2007). Table 2
shows the Pearson correlation coefficients relating categorization of words to that of environmental sounds for
each group within each compression speed and overlap
condition. For example, performance for the words in
the nonoverlapping, slow-compression condition was compared with performance for the environmental sounds in
the same overlap and compression condition. Asterisks
indicate significant one-tailed correlations (p , 0.05). A
few correlations were revealed for some groups across conditions; however, the most striking results were those of
the adults with HL. Listeners who were able to categorize
words well also appeared to do well with environmental
sounds, whereas those who did poorly with words also
tended to do poorly with the environmental sounds. The
relationship between speech and environmental sound
performance was significant in every condition for the
adults with HL whereas it occurred only in a few conditions for all other groups.
843
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Journal of the American Academy of Audiology/Volume 25, Number 9, 2014
Table 2. Pearson Correlation Coefficients Relating Performance for Word Categorization to Performance for
Environmental Sounds within Each Processing Speed
Nonoverlapping
Age Group
Adults
Adults
Children
Children
Overlapping
Hearing Group
Slow
Fast
Adaptive
Slow
Fast
Adaptive
NH
HL
NH
HL
–0.22
0.53*
0.51*
0.43
–0.01
0.64*
0.26
0.48*
0.39*
0.58*
0.29
0.50*
–0.10
0.56*
0.44*
0.13
–0.07
0.49*
0.20
0.42
0.09
0.59*
0.27
0.25
*Significant one-tailed correlations (p , 0.05).
DISCUSSION
T
he goal of this study was to examine the impact of
amplitude compression on children’s and adults’
ability to categorize words and environmental sounds
in complex listening environments. Three forms of
amplitude compression (fast, slow, and adaptive) were
simulated in complex acoustic trials containing a triggering stimulus, a target stimulus, and a low-level playground noise. The trigger preceded the target or
overlapped with it such that the temporal effects of compression were expected to be most deleterious to the
perception of the partial stimuli. Overall, the performance of the listeners with HL was poorer than that of the
listeners with NH with unique patterns of performance
emerging with respect to age, stimulus overlap, and
compression speed for both groups.
For example, significant effects of age were observed
for the listeners with NH but not for the listeners with
HL. Typically, children with HL demonstrate the most
difficulty with auditory tasks; however, in this study
their performance was equivalent to that of the adults
with HL. This unusual finding suggests that either the
children with HL performed better than expected or the
adults performed more poorly. It is possible that the children were better suited to the dual-task nature of the
paradigm than the adults; however, it is also possible
that the cognitive skills of some of the adults with
HL were in decline, resulting in performance similar
to that of children with developing cognitive skills. This
finding is supported by the significant correlations
between performance for the speech and environmental
sounds for the adults with HL in every listening condition. We know that adults with HL who have poor
cognitive skills are less able to tolerate the subtle distortions to speech that occur with fast compression
(Gatehouse et al, 2006; Souza et al, 2000; Jenstad
and Souza, 2005, 2007; Cox and Xu, 2010; Lunner
and Sundewall-Thorén, 2007), and in fact, the adults
with HL demonstrated their poorest performance in
the fast compression condition. Given that significant
correlations were found in every listening condition,
it is possible that the effects of declining cognitive skills
affected performance in each condition. However, few
correlations between the perception of speech and environmental sounds were found for the children with HL,
suggesting that the mechanisms that govern these processes may differ in children relative to adults and are
indeed a question for further research.
Another example of the differences between the NH
and HL groups was that the listeners with NH performed best with slow compression whereas the listeners with HL performed best with adaptive compression.
These results suggest that the subtle effects of compression speed interact with healthy and impaired ears in a
unique fashion. Why the listeners with NH favored
slow-acting compression is unknown and would also
be an interesting subject for further research. In terms
of the practical application of these results, it is important to bear in mind that, although significant,
the performance of the listeners with NH varied by only
a few percentage points across compression conditions
whereas larger variations in performance were observed
for the listeners with HL (as much as 20 percentage
points). For these listeners, the different types of subtle
distortion caused by slow and fast compression reduced
their performance to similarly low levels. These findings are consistent with previous studies showing little
differences in speech perception between fast- and slowcompression conditions (Moore et al, 2004, 2010; Shi
and Doherty, 2008). Not only does this suggest that listeners with HL would benefit from amplification with
adaptive compression, the results support the argument that amplification decisions for adults and children with HL should not be based on the results of
listeners with NH.
Finally, in the nonoverlapping condition, the performance of the listeners with HL was nearly as good
as that of the listeners with NH for the speech stimuli
but not for the environmental sounds. Although it
is fairly clear that similar cognitive mechanisms are
involved in the perception of speech and other acoustic
stimuli (Fowler and Rosenblum, 1990), the results of
the present study suggest that speech may be prioritized among individuals with HL. That is, adults and
children with HL may consider environmental sounds
as a source of unwanted noise that may prevent them
from interacting with these sounds accurately and
844
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Fast, Slow, and Adaptive Amplitude Compression/Pittman et al
effectively (Banerjee, 2011). In fact, informal interactions with several of the adults with HL at the conclusion
of testing suggest that they avoid environmental sounds
whenever possible to optimize their speech perception.
Such a strategy may be detrimental to both children
and adults with HL in that it may limit their social, educational, and vocational interactions as well as their ability to dynamically interact with their environment.
Efforts to make environmental sounds accessible to individuals with HL may be of significant benefit to them.
Limitations of the Present Study
There were a few limitations of the present study.
First, and most important, the results should be generalized to hearing aid users with caution. Because the
three compression conditions examined could not be
generated using a single, wearable device, it was necessary to simulate them through digital processing. Thus,
the listeners with HL did not receive amplification with
compression characteristics consistent with current
multichannel, multistage compression devices. Also,
the stimuli do not represent all complex listening environments. Thus, the results confirm the potential for
benefit from devices using adaptive compression parameters compared with devices using fast or slow compression parameters.
Second, the stimuli were not counterbalanced across
compression conditions. The complexity of the stimuli
required 216 trials to be created by combining 432 stimuli. To counterbalance the stimuli across compression
conditions, a total of 1,296 stimuli would have been
required, which was not feasible during the development of the study. It is possible that list effects were
present (one list being easier than the others); however,
little evidence of such effects can be observed in the
results. For example, the better word categorization
of the listeners with HL in the adaptive compression
condition was not observed in the listeners with NH
for the same stimuli. Likewise, the better word categorization for the slow compression condition by the listeners
with NH was not observed for the listeners with HL.
Whether or not list effects are present, the complexity
of the stimuli alone warrants replication of the results
with similar or related stimuli to confirm, and possibly
identify, further benefits of adaptive compression.
Third, the source of the words and environmental sounds
differed substantively. That is, a single sound source
(talker) was used for the speech stimuli whereas the environmental sounds were gathered from multiple sources.
Also, the speech stimuli were recorded in quiet but were
presented in noise and lacked the vocal effort that would
be expected in such surroundings (Pittman and Wiley,
2001). It is recommended that future research in this area
include speech samples from a wide range of talkers produced in the same or similar listening environments.
Finally, the categorization task for the environmental sounds was uncharacteristic of how a listener would
respond to these sounds. A clever experiment might
require a listener to respond dynamically to both speech
and environmental sounds in some fashion.
CONCLUSIONS
A
mplitude compression speed imposes subtle alterations to the acoustic signal that can affect the performance of both children and adults with HL. Although
fast, slow, and adaptive compression parameters were
simulated in the present study, the results suggest that
the performance of listeners with HL may be reduced
similarly when fast or slow amplitude compression is
used. On the other hand, listeners with HL may derive
significant benefit from hearing instruments that use
adaptive amplitude compression, especially in complex
listening environments.
Acknowledgments. The authors thank the staff of the
Pediatric Amplification Laboratory for their assistance with
the data collection (Rachel Henrickson, Tia Mulrooney, and
Amanda Willman), Fares El-Azm and Jan-Mark de Haan
for their help with stimulus preparation, and Thomas Behrens
and Kamilla Angelo for sharing their expertise in this area of
hearing aid research. Appreciation also goes to the participants
who shared their precious time with us during this project.
REFERENCES
American National Standards Institute. (1997). Methods for Calculation of the Speech Intelligibility Index (ANSI S3.5-1997). New
York, NY: American National Standards Institute.
American National Standards Institute. (1999) Maximum Permissible Ambient Noise Levels for Audiometric Test Rooms (ANSI S3.1
1-1999). New York, NY: American National Standards Institute.
Banerjee S. (2011) Hearing aids in the real world: typical automatic behavior of expansion, directionality, and noise management. J Am Acad Audiol 22(1):34–48.
Boike KT, Souza PE. (2000) Effect of compression ratio on
speech recognition and speech-quality ratings with wide dynamic
range compression amplification. J Speech Lang Hear Res 43(2):
456–468.
Cox RM, Xu J. (2010) Short and long compression release times:
speech understanding, real-world preferences, and association
with cognitive ability. J Am Acad Audiol 21(2):121–138.
Davies-Venn E, Souza P, Brennan M, Stecker GC. (2009) Effects of
audibility and multichannel wide dynamic range compression on
consonant recognition for listeners with severe hearing loss. Ear
Hear 30(5):494–504.
Dawson PW, Decker JA, Psarros CE. (2004) Optimizing dynamic
range in children using the nucleus cochlear implant. Ear Hear 25
(3):230–241.
Dockrell JE, Shield B. (2004) Children’s perceptions of their acoustic environment at school and at home. J Acoust Soc Am 115(6):
2964–2973.
845
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Journal of the American Academy of Audiology/Volume 25, Number 9, 2014
Fabiani M, Kazmerski VA, Cycowicz YM, Friedman D. (1996)
Naming norms for brief environmental sounds: effects of age
and dementia. Psychophysiology 33(4):462–475.
Moore BC, Stainsby TH, Alcántara JI, Kühnel V. (2004) The effect
on speech intelligibility of varying compression time constants in a
digital hearing aid. Int J Audiol 43(7):399–409.
Fowler CA, Rosenblum LD. (1990) Duplex perception: a comparison of monosyllables and slamming doors. J Exp Psychol Hum
Percept Perform 16(4):742–754.
Naylor G, Johannesson RB. (2009) Long-term signal-to-noise ratio
at the input and output of amplitude-compression systems. J Am
Acad Audiol 20(3):161–171.
Gatehouse S, Naylor G, Elberling C. (2006) Linear and nonlinear
hearing aid fittings–2. Patterns of candidature. Int J Audiol 45(3):
153–171.
Pittman A. (2011) Children’s performance in complex listening
conditions: effects of hearing loss and digital noise reduction.
J Speech Lang Hear Res 54(4):1224–1239.
Gygi B, Shafiro V. (2007) General functions and specific applications of environmental sound research. Front Biosci 12:
3152–3166.
Pittman AL, Wiley TL. (2001) Recognition of speech produced in
noise. J Speech Lang Hear Res 44(3):487–496.
Hornsby BW, Ricketts TA. (2001) The effects of compression ratio,
signal-to-noise ratio, and level on speech recognition in normalhearing listeners. J Acoust Soc Am 109(6):2964–2973.
Jenstad LM, Souza PE. (2005) Quantifying the effect of compression hearing aid release time on speech acoustics and intelligibility. J Speech Lang Hear Res 48(3):651–667.
Jenstad LM, Souza PE. (2007) Temporal envelope changes of compression and speech rate: combined effects on recognition for older
adults. J Speech Lang Hear Res 50(5):1123–1138.
Johnson JA, Cox RM, Alexander GC. (2010) Development of
APHAB norms for WDRC hearing aids and comparisons with original norms. Ear Hear 31(1):47–55.
Kam AC, Wong LL. (1999) Comparison of performance with wide
dynamic range compression and linear amplification. J Am Acad
Audiol 10(8):445–457.
Kim M, Chang SI, Seong JC, et al. (2012) Road traffic noise: annoyance, sleep disturbance, and public health implications. Am J Prev
Med 43(4):353–360.
Ranjbar P, Borg E, Philipson L, Stranneby D. (2008) Auditive
identification of signal-processed environmental sounds: monitoring the environment. Int J Audiol 47(12):724–736.
Rudner M, Rönnberg J, Lunner T. (2011) Working memory supports listening in noise for persons with hearing impairment.
J Am Acad Audiol 22(3):156–167.
Seewald RC, Cornelisse LE, Ramji KV, Sinclair ST, Moodie KS,
Jamieson DG. (1997). DSL v4.1 for Windows: A software Implementation of the Desired Sensation Level (DSL[i/o]) Method for
Fitting Linear Gain and Wide-Dynamic-Range Compression Hearing Instruments. User’s manual. (Version 4.1) [Computer software]. London, ON: The University of Western Ontario Hearing
Health Care Research Unit.
Shafiro V. (2008) Development of a large-item environmental
sound test and the effects of short-term training with spectrally-degraded stimuli. Ear Hear 29(5):775–790.
Shafiro V, Gygi B. (2004) How to select stimuli for environmental
sound research and where to find them. Behav Res Methods Instrum Comput 36(4):590–598.
Kraft RH. (1982) Relationship of ear specialization to degree of
task difficulty, sex, and lateral preference. Percept Mot Skills
54(3):703–714.
Shepherd D, Welch D, Dirks KN, Mathews R. (2010) Exploring the
relationship between noise sensitivity, annoyance and healthrelated quality of life in a sample of adults exposed to environmental noise. Int J Environ Res Public Health 7(10):3579–3594.
Kraft RH, Harper LV, Nickel LD. (1995) Stability and ear advantage for dichotically presented environmental sound stimuli:
development during early childhood. Percept Mot Skills 80(2):
611–624.
Shi LF, Doherty KA. (2008) Subjective and objective effects of fast
and slow compression on the perception of reverberant speech in
listeners with hearing loss. J Speech Lang Hear Res 51(5):
1328–1340.
Larson VD, Williams DW, Henderson WG, et al.; NIDCD/VA
Hearing Aid Clinical Trial Group. (2000) Efficacy of 3 commonly
used hearing aid circuits: A crossover trial. JAMA 284(14):
1806–1813.
Smith AW. (1998) The World Health Organisation and the prevention of deafness and hearing impairment caused by noise. Noise
Health 1(1):6–12.
Looi V, Arnephy J. (2010) Environmental sound perception of
cochlear implant users. Cochlear Implants Int 11(4):203–227.
Lunner T, Sundewall-Thorén E. (2007) Interactions between cognition, compression, and listening conditions: effects on speech-innoise performance in a two-channel hearing aid. J Am Acad
Audiol 18(7):604–617.
McFadden B, Pittman A. (2008) Effect of minimal hearing loss on
children’s ability to multitask in quiet and in noise. Lang Speech
Hear Serv Sch 39(3):342–351.
Souza PE. (2002) Effects of compression on speech acoustics, intelligibility, and sound quality. Trends Hear 6(4):131–165.
Souza PE, Jenstad LM, Folino R. (2005) Using multichannel widedynamic range compression in severely hearing-impaired listeners: effects on speech recognition and quality. Ear Hear 26(2):
120–131.
Souza PE, Turner CW. (1999) Quantifying the contribution of
audibility to recognition of compression-amplified speech. Ear
Hear 20(1):12–20.
Moe AH, Hopkins CJ, Rush RT. (1982) The Vocabulary of First
Grade Children. Springfield, IL: Thomas Publishers.
Souza PE, Yueh B, Sarubbi M, Loovis CF. (2000) Fitting hearing
aids with the Articulation Index: impact on hearing aid effectiveness. J Rehabil Res Dev 37(4):473–481.
Moore BC, Füllgrabe C, Stone MA. (2010) Effect of spatial separation, extended bandwidth, and compression speed on intelligibility in a competing-speech task. J Acoust Soc Am 128(1):
360–371.
Spaulding TJ, Plante E, Vance R. (2008) Sustained selective attention skills of preschool children with specific language impairment: evidence for separate attentional capacities. J Speech
Lang Hear Res 51(1):16–34.
846
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53
Fast, Slow, and Adaptive Amplitude Compression/Pittman et al
Stelmachowicz PG, Hoover BM, Lewis DE, Kortekaas RW,
Pittman AL. (2000) The relation between stimulus context, speech
audibility, and perception for normal-hearing and hearing-impaired
children. J Speech Lang Hear Res 43(4):902–914.
Stone MA, Moore BC. (2003) Effect of the speed of a single-channel
dynamic range compressor on intelligibility in a competing speech
task. J Acoust Soc Am 114(2):1023–1034.
Tyler RS. (1994) Advantages of disadvantages expected and
reported by cochlear implant patients. Am J Otol 15(4):523–531.
van Lancker D, Cornelius C, Kreiman J, Tonick I, Tanguay P,
Schulman ML. (1988) Recognition of environmental sounds in
autistic children. J Am Acad Child Adolesc Psychiatry 27(4):
423–427.
Studebaker GA. (1985) A “rationalized” arcsine transform. J Speech
Hear Res 28(3):455–462.
Walden BE, Surr RK, Cord MT, Edwards B, Olson L. (2000) Comparison of benefits provided by different hearing aid technologies.
J Am Acad Audiol 11(10):540–560.
Thawin C, Kanchanalarp C, Lertsukprasert K, Cheewaruangroj W,
Khantapasuantara K, Ruencharoen S. (2006) Auditory performance of cochlear implant children aged 2-5 years. J Med Assoc Thai
89(11):1923–1927.
Zhao F, Stephens SD, Sim SW, Meredith R. (1997) The use of
qualitative questionnaires in patients having and being considered for cochlear implants. Clin Otolaryngol Allied Sci 22(3):
254–259.
Appendix.
Speech and Environmental Sounds Stimuli by Category
Speech
Environmental Sounds
Animal
Food
Person
Mechanical
Biological
Animal
alligator
bat
bear
buzzard
camel
cat
chipmunk
cow
crocodile
deer
dolphin
donkey
eagle
elephant
fox
frog
giraffe
goldfish
gorilla
hamster
horse
lion
mouse
octopus
owl
penguin
porcupine
rabbit
raccoon
seal
squirrel
toad
walrus
weasel
whale
zebra
apple
bacon
banana
cabbage
cake
candy
carrots
cheese
cherry
chocolate
corn
cracker
cream
donut
fudge
grape
hamburger
hotdog
jelly
lollipop
macaroni
meatloaf
milk
noodle
oatmeal
pancake
pineapple
pizza
popcorn
popsicle
potato
sausage
soup
strawberry
sugar
tomato
astronaut
baby
boy
brother
burglar
captain
carpenter
clown
cousin
cowboy
dentist
detective
doctor
father
friend
girl
girlfriend
grandma
grandpa
janitor
king
lifeguard
mailman
man
pilot
president
priest
prince
queen
reporter
robber
sailor
sister
uncle
wife
woman
alarm1
alarm2
camera1
camera2
carhonk1
carhonk2
door1
door2
doorbell
garbage can
helicopter1
helicopter2
jar opening
lawnmower
phone1
phone2
pop1
pop2
pop3
power tool1
power tool2
power tool3
school bell
scrape1
scrape2
scrape3
siren1
siren2
squeak1
squeak2
squeak3
tool1
tool2
train
trash lid1
trash lid2
blow nose1
blow nose2
burp1
burp2
burp3
clear throat
cough1
cough2
cough3
cough4
cry1
cry2
footsteps
gargle
gasp1
gasp2
gasp3
grunt1
grunt2
grunt3
hiccups
kiss1
kiss2
laugh1
laugh2
laugh3
scream1
scream2
sigh
slurp
sneeze1
sneeze2
sneeze3
whistle1
whistle2
yawn
alligator
bear1
bear2
bear3
bear4
bird1
bird2
bird3
bird4
bird5
cat1
cat2
chicken1
chicken2
cow
dog1
dog2
donkey
frog1
frog2
horse1
horse2
horse3
lamb1
lamb2
monkey1
monkey2
monkey3
monkey4
owl1
owl2
owl3
pig
rat
seal
sheep
847
Delivered by Ingenta to: Arizona State University
IP : 149.169.180.108 On: Fri, 21 Nov 2014 18:27:53