Effects Of Extended Bandwidth

WIDEXPRESS no.24
APRIL 2010

Review article:
By Patricia Stelmachowicz, Ph.d

ExtEndIng HEARIng AId BAndwIdtH:
EffEctS, challEngES And SolutionS

Boothroyd and Medwetsky (1992) were the first to EffEct of bandwidth on spEEch pErcEption
suggest that the bandwidth of hearing aids is nar- It has been demonstrated that children with hearing
rower than that of the acoustic speech signal. Earlier loss often omit /s/ in final position in speech production
studies related to telephone communication (French (Elfenbein, Hardin-Jones, and davis, 1994). Kortekaas
and Steinberg, 1947; Fletcher, 1953) had found that and Stelmachowicz (2000) examined developmental
an upper limit of 3000 Hz was sufficient to support effects in the perception of /s/ and /z/ when used as
speech perception for adults with normal hearing. This an inflectional morpheme (denoting tense, plurality,
information, as well as the many technical challenges possession). Participants were 5, 7, and 10-year old
associated with bandwidth extension, has contributed children and adults with normal hearing. The upper
to the slow progress in this area. In their 1992 study, bandwidth was varied adaptively to construct psycho-
Boothroyd and Medwetsky focused on the acoustic metric functions. Results revealed significant differ-
characteristics of /s/ spoken by adult male and female ences between the adults and each of the younger age
talkers in an effort to determine the optimal upper groups, but no differences among the three groups of
bandwidth for hearing aids. Their results revealed that children. clarity ratings for target words in low-pass fil-
the lowest frequency prominent spectral peak of /s/ tered sentences revealed increased clarity as a function
varies both within and across talkers and ranges from of bandwidth for all groups.
2.9 kHz to 8.9 kHz. They concluded that an upper fre-
quency limit of 10 kHz would be needed to ensure the In 2001, Stelmachowicz et al. evaluated the effects of
audibility of /s/ for most talkers. stimulus bandwidth on the perception of /s/ by normal-
hearing (NH) and hearing-impaired (HI) adults and
Since the majority of hearing-aid users are adults who children (5-8 year olds). Fig. 1 shows the low-pass filter
acquired hearing loss later in life, lexical, syntactic, and frequency at which performance plateaus as a func-
semantic information can be used to “fill in the blanks” tion of talker for the four groups of participants. when
when processing speech with a restricted bandwidth. listening to a male talker, NH adults reach maximum
This may be one reason why the acoustic needs of performance with a bandwidth of 4 kHz, but the re-
children with hearing loss were not readily considered maining three groups required a bandwidth of 5 kHz.
in the design of hearing instruments for many years. A bandwidth of 9 kHz was needed to reach maximum

www.wIdEx.coM wIdExPRESS 1

from 2-4 kHz was most important for the perception
of these phonemes spoken by a male talker, but that a
wider frequency range (2-8 kHz) was needed for the
female talker.

Pittman et al. (2003) examined the spectral character-
istics of speech produced by adults and 2-4 year old
children measured at both 0 degrees azimuth and at
the hearing-aid microphone. The goal was to quantify
the audibility of specific phonemes for the purpose
of monitoring one’s own speech. Fig. 2 illustrates the
mean short-term spectra of /s/ and /∫/ (produced by
adult males, adult females, and 2-4 year old children)
measured at the hearing-aid microphone. For both
phonemes, stimulus amplitude is similar for the male
and female talkers, but greater mid-frequency energy
is apparent for the male talker. overall amplitude is
markedly reduced for the child talkers; this result may
be due to reduced vocal effort by young children and
suggests that the ability to adequately monitor one’s
own productions may be compromised in this group.

fig. 1. Low-pass filter frequency for maximum performance as
a function of talker. data are shown for four groups of listeners
(Reproduced with permission, Stelmachowicz et al., 2001).

performance for the perception of /s/ spoken by female
and child talkers for all listeners.
In a follow-up study, Stelmachowicz et al. (2002)
examined perception of /s/ and /z/ spoken by male and
female talkers for 36 NH children (3-5 years) and 40 fig. 2. Spectrum level as a function of frequency for /s/ and /∫/
produced by a male, female, and child talker (Reproduced with
HI children (5-13 years). The latter group wore their permission, Pittman et al., 2003).
personal hearing aids. By age 5 years, 3 months mean
scores for both the female and male talkers were > 95%
for the children with NH. Results for the HI children
revealed an improvement in performance as a function
of age, with greater variability for the youngest chil-
dren. collapsed across age, slightly better mean scores
were observed for the male talker (90%) compared to
the female talker (85%). Additional analyses of the HI
data revealed that audibility in the frequency region


Early work in the area of hearing-aid bandwidth EffEct of bandwidth on spEEch
focused on the perception and production of frica- dEvElopmEnt
tives. To determine if a restricted bandwidth might also Interestingly, the results of a longitudinal study of
influence more complex auditory skills, Stelmachowicz phonetic development in infants with hearing loss who
et al. (2007) examined the effects of bandwidth on were identified early and fitted with hearing aids by
word and fricative perception, novel-word learning, 4.7 months (on average) and age-matched infants with
and listening effort in both NH and HI children. Results normal hearing revealed a significant delay in fricative
revealed significant effects of bandwidth on perception and affricate development in the children with hear-
of the fricatives /s/, /z/, and /ð/ (Fig. 3) and monosyl- ing loss (Moeller et al., 2007). The left panel of Fig. 5
labic words (Fig. 4). However, no significant effects of illustrates that the non-fricative sounds are delayed in
bandwidth were found for either novel word learning or the HI group relative to the NH children, but that the
listening effort. pattern of development over time is similar for the
two groups. For fricatives and affricates (right panel),
however, the gap between the two groups increases
as a function of age. It is likely that reduced audibility
of these speech sounds due to a restricted hearing-
aid bandwidth, as well as the common occurrence of
reverberation and noise, may be contributing factors
to these delays in phonological development. Fig.6
illustrates the development of three classes of speech
sounds for the HI children in relation to the NH group.
For example, at 16 months of age, the HI children are at
ð ð
80% of the NH group in terms of vowel acquisition and
by 24 months, they have narrowed the gap to 90%. A
similar developmental trend is observed for the stops,
fig. 3. Percent correct phoneme recognition for /s/, /z/, and /ð/
nasals, glides, and liquids. consistent with the data re-
for listeners with normal hearing and hearing loss. (Reproduced
with permission, Stelmachowicz et al., 2007). ported above, a very different pattern emerges for the
fricative class where overall performance is at 40% of
the NH group at 16 months and has actually decreased
to 30% of normal by 24 months.

fig. 4. Percent correct word scores for listeners with hearing
fig 5. Percent sounds produced as a function of age group for
loss. (Reproduced with permission, Stelmachowicz et al., 2007).
non-fricatives and fricatives/affricates. data are shown for listen-
ers with and without hearing loss. (Reproduced with permission,
Moeller et al., 2007).

EffEct of hEaring-aid bandwidth on
quality judgmEnts for spEEch and music
Recent studies also have examined the effects of
stimulus bandwidth on the perceived quality of speech
and/or music by adults with hearing loss. Moore and
Tan (2003) investigated the perceived naturalness of
speech (spoken by male and female talkers). They re-


coupling for BTE instruments (i.e., tubing and earmold)
essentially behaves like a low-pass filter. while the
bandwidth of ITE and receiver-in-the-ear-canal devices
can be wider than BTE instruments, ear canal size and
concerns for hearing-aid retention typically prohibit the
use of these devices with infants and young children.
Finally, acoustic feedback is likely to increase with an
expansion of the upper frequency limit of hearing in-
struments. This is especially an issue with the pediatric
population, where the fast growth of the outer ear
makes frequent ear impressions necessary, to ensure a
tight fitting earmould. Solutions need to be developed
to address the above mentioned challenges.

potEntial solutions
The most obvious solution to the limited bandwidth
problem is to develop methods to extend the band-
width of hearing aids without incurring new complica-
fig. 6. Proportion of vowels, stops, nasals, glides and liquids and
tions. An alternative approach, however, is the use of
fricatives produced correctly by children with hearing loss at 16
and 24 months of age. (Reproduced with permission, Moeller et frequency-lowering schemes. Several forms of this
al., 2007). technology are currently available in wearable hear-
ing instruments. with these devices, real-time digital
analysis is used to “shift” or “compress” high frequency
ported a marked degradation in performance when the sounds to lower frequencies to improve auditory ac-
upper cutoff frequency was decreased below 10,869 cess. In cases where hearing loss in the high frequen-
Hz. For music, the highest quality ratings occurred at cies is severe and/or “dead regions” exist, a widened
an upper cutoff of 16,854 Hz. More recently, Ricketts et bandwidth will be unlikely to provide sufficient audibil-
al., (2008) studied the effects of bandwidth on sound ity of speech. Frequency-lowering provides an alterna-
quality for adults with hearing loss. Paired comparisons tive and may be the only option in cases where extend-
were used to determine preferred sound quality for ing the bandwidth does not achieve audibility.
music and for a movie sound track. Results revealed
that preferences were related to the slope of hearing nEw solutions bring nEw challEngEs
loss in the 4-12 kHz region. Specifically, individuals with As manufacturers strive to extend the upper band-
steeply-sloping hearing losses showed a preference for width hearing aids, an additional challenge emerges.
narrower bandwidths. To date, no similar studies have Specifically, for frequencies above 4 kHz, it is known
been conducted in children with hearing loss. that standing waves will occur in the ear canal due to
an interaction between the incoming sound and reflec-
Expanding thE bandwidth of hEaring aids: tions from the tympanic membrane. during probe-mi-
tEchnological challEngEs crophone measures, these minima will be observed as
despite major advances in hearing-aid technology over notches in the aided frequency response (wiener and
the past decade, the effective bandwidth of current Ross, 1946; Khanna and Stinson, 1985). These notches
devices typically is narrower than that of the acoustic are artifacts of the measurement process and are not
speech signal. The current ANSI standard for determi- believed to be indicative of the input to the middle
nation of the frequency range (bandwidth) of hearing ear and cochlea. In the region where aided gain ap-
aid can tend to overestimate the usable bandwidth pears to be insufficient, clinicians are likely to increase
of hearing aids. Specifically, even in cases where the hearing-aid gain to compensate for what is thought to
upper cutoff frequency is relatively high (e.g., 6-7 be inadequate audibility. These increases in gain may
kHz), the actual use gain may be insufficient to pro- result in excessive amplification with a subsequent risk
vide audibility for the high-frequency components of to residual hearing (Hawkins, 1982). with significant
speech. This should also be seen in conjunction with increases in hearing-aid bandwidth, valid and reliable
the dynamic capabilities of the hearing aid receiver in techniques to measure sound levels in the ear canal
this frequency region. must be developed (Mccreery et al. 2008; Lewis et al.,
There are multiple reasons why an expansion of 2009).
bandwidth is technologically challenging. It is difficult
to build wideband transducers that can provide both
high gain and low distortion. In addition, the acoustic


references Ricketts, T. A., dittberner, A. B., & Johnson, E. E.
(2008). High-frequency amplification and sound qual-
Boothroyd, A. and Medwetsky, L. (1992). Spectral dis- ity in listeners with normal through moderate hearing
tribution of /s/ and the frequency response of hearing loss. J Speech Lang Hear Res, 51, 160-172.
aids. Ear and Hearing, 13, 150-157.
Stelmachowicz, P., Pittman, A., Hoover, B., and Lewis,
Elfenbein, J.L., Hardin-Jones, M.A., and davis, J.M. d. (2001). The effect of stimulus bandwidth on the per-
(1994). oral communication skills of children who are ception of /s/ in normal and hearing-impaired children
hard of hearing. J Speech Lang Hear Res, 37, 216-226. and adults. J Acoust Soc Am, 110, 2183-2190.

Fletcher, H. (1953). Speech and Hearing in communica- Stelmachowicz, P.G., Pittman, A.L., Hoover, B.M., &
tion. New York: Van Nostrand. Lewis, d.L. (2002). Aided perception of /s/ and /z/ by
hearing-impaired children. Ear and Hearing, 23, 316-324.
French, N.R. and Steinberg, J.c. (1947). Factors govern-
ing the intelligibility of speech sounds. J Acoust Soc Stelmachowicz, P.G., Lewis, d.E., choi, S., Hoover, B.
Am, 19, 90-119. (2007). Effect of stimulus bandwidth on auditory skills
in normal-hearing and hearing-impaired children. Ear
Hawkins, d. B. (1982). overamplification: A well-docu- and Hearing, 28, 483-494.
mented case report. J Speech Hear Dis. 47, 382-384.
weiner, F. M., & Ross, d. A. (1946). The pressure dis-
Khanna, S. M., & Stinson, M. R. (1985). Specification of tribution in the auditory canal in a progressive sound
the acoustical input to the ear at high frequencies. J field. J Acoust Soc Am, 18, 401-408.
Acoust Soc Am, 77, 577-589.

Kortekaas, R. and Stelmachowicz, P.G. (2000). Band-
width effects on children’s perception of the inflection-
al morpheme /s/: Acoustical measurements, auditory
detection, and clarity rating. J Speech Lang Hear Res,
43, 645-660.

Lewis, J. et al., (2009). comparison of in-situ calibra-
tion methods for quantifying ear-canal sound level.
American Auditory Society Meeting. Scottsdale, AZ,
March.

Mccreery, R., et al. (2008). Use of forward pres-
sure level (FPL) to minimize the influence of acoustic
standing waves during probe-microphone hearing-aid
measures. International Hearing Aid conference. Lake
Tahoe, cA, August.

Moeller, M.P., Hoover, B., Putman, c., Arbataitis, K.,
Bohnenkamp, G., Peterson, B., wood, S.L., Lewis, d.E.,
Pittman, A.L. & Stelmachowicz, P.G. (2007). Vocaliza-
tions of infants with hearing loss compared to infants
with normal hearing - Part I: Phonetic development. Ear
and Hearing. 28: 605-627.

Moore, B. c., & Tan, c. T. (2003). Perceived naturalness
of spectrally distorted speech and music. J Acoust Soc
Am, 114, 408-419.

Pittman, A.L., Stelmachowicz, P.G., Lewis, d.E., &
Hoover, B.M. (2003). Spectral characteristics of speech
at the ear: Implications for amplification in children. J
Speech Lang Hear Res, 46, 649-657.

Printed by KLS / 2010-06
9 502 1969 001 #01

Effects Of Extended Bandwidth

More Related Content

Similar to Effects Of Extended Bandwidth (20)

More from Geoffrey Cooling (20)

Effects Of Extended Bandwidth