Minerva Access is the Institutional Repository of The University of Melbourne
Author/s:
van Hoesel, Richard J. M.; Clark, Graeme M.
Title:
Speech results with a bilateral multi-channel cochlear implant subject for spatially separated
signal and noise
Date:
1999
Citation:
van Hoesel, R. J. M., & Clark, G. M. (1999). Speech results with a bilateral multi-channel
cochlear implant subject for spatially separated signal and noise. Australian Journal of
Audiology, 21(1), 23-28.
Persistent Link:
http://hdl.handle.net/11343/27537
File Description:
Speech results with a bilateral multi-channel cochlear implant subject for spatially separated
signal and noise
Terms and Conditions:
Terms and Conditions: Copyright in works deposited in Minerva Access is retained by the
copyright owner. The work may not be altered without permission from the copyright owner.
Readers may only download, print and save electronic copies of whole works for their own
personal non-commercial use. Any use that exceeds these limits requires permission from
the copyright owner. Attribution is essential when quoting or paraphrasing from these works.
c=---
OTOPUBS #1286-==:J
Speech Results with a Bilateral Multi­
Channel Cochlear Implant Subject for
Spatially Separated Signal and Noise
RICHARD J.M. van HOESEL AND G.M. CLARK
Cooperative Research Centre for Cochlear Implant, Speech and Hearing Research
Speech tests in noise were administered to a bilater­
ally implanted cochlear implant subjecl. Performance
for simultaneous use of two identical implants, with
the same speech processing strategy on two indepen­
dent standard clinical processors, was compared with
that of the beller performing monaural side alone.
Speech was presented at an angle of 45 degrees
toward one ear, with noise at 45 degrees toward the
contralateral sidc. Tests were also administered for
speech and noise reversed in location. When the
speech signal was on the same side as the subject's
beller performing ear, monaural and binaural tests
resulted in similar scores. When the speech was on
the opposite side, however, the binaural condition
showed significantly beller speech scores. The resulls
indicate that binaural implants can provide improved
performance in noise when speech and noise arc
spatially separated.
Binaural prosthesis with multi-channel
implants has been studied in our laboratory
since bilateral implantation of our first
subject, PI, in 1989. At that time the clinical
speech strategy in use for the Cochlear cr­
22M system was the "FO/Fl/F2" strategy
(Blarney et. aI., 1987). Using that strategy,
PI demonstrated a binaural advantage
compared to his better performing monaural
side alone when speech and noise were
presented from the same source (van Hoesel
et aI., 1993). This result encouraged bilateral
implantation of a second subject, P2, in
1993. The "Spectral Maxima Sound Proces­
sor" (Mc'Kay et aI., 1992) was then showing
improved performance over the FO/Fl/F2
strategy, and was therefore also adopted for
the binaural subjects. Speech tests, again
using spatially coincident speech and noise,
showed that the better side alone performed
at least as well as the binaural case for both
subjects (van Hoesel, 1998). Psychophysical
studies conducted with both subjects further­
more indicated that inter-aural time delays
were not well-perceived for both subjects
when compared to the normal hearing case
(van Hoesel and Clark, 1997). This implied
that binaural unmasking, which has been
shown to improve intelligibility in noise for
normal hearing (Licklider, 1948, Hirsh,
1950), might not be available to binaural
implant subjects. The work presented here,
however, re-exam ines the possibility that
bilateral implantation can offer advantages in
noise even in the absence of binaural process­
ing per se, simply by attending the ear that
has the better signal to noise ratio (SIN). This
has also been a proposed mechanism explain­
ing binaural intelligibility gain in normal
hearing subjects (e.g. Zurek, 1983).
METHODS
Patient details
At present, P2 is the only bilateral implant
subject available in Melbourne, Australia, so
that the results presented here are for this
subject alone. This subject was implanted
monaurally in 1985 with a Cochlear Cr-22M
device after suffering a profound hearing loss
Correspondence and reprint requests Richard JM. van Hoesel and GM Clark, Cooperative Research Centre for Cochlear
implant. Speech and Hearing Research, 384-388 Albert St East Melbourne VIC 3002, Australia
THE AUSTRALIAN JOURNAL OF AUDIOLOGY
VOLUME 21 NUMBER 1 MAY 1999 pp 23-28
23
RICHARD J.M van HOESEL AND G.M CLARK
due to Meniere's disease. He was fitted with a
second CI-22M device in 1993. X-ray data
from this subject, obtained using a modified
Stenvers' view (Marsh et a!., 1993), as well as
psychophysical data (van Hoesel and Clark,
1997) indicated that the offset in insertion
depth between the two sides was approxi­
mately 8.5 electrode bands (about 6.5 mm).
Earlier speech tests with this subject compared
binaural performance for speech maps that
made allowance for the offset between the two
sides with those that did not, and showed no
significant difference between the two condi­
tions (van Hoese1, 1998). For this reason, and
given that on the side with the shallower inser­
tion only about 10 bands were available for
stimulation, for the present study it was
decided not to compensate for the electrode
insertion offset between the two sides.
Results for P2 with an earlier binaural
speech processor, in 1994, for open set SIT
sentences (Whitford et. a!., 1995) in quiet,
showed monaural scores of about 60% and
20% for the left and right sides respectively
(van Hoesel, 1998). At SIN levels of 5 dB
when using the right side alone, performance
dropped to 0%. This implied that, for the
present study, testing near 5 dB SIN with the
monaural side that performed worse might be
very frustrating for the subject. Furthermore,
since the work described here was primarily
concerned with exploring whether a binaural
advantage was possible with respect to the
better side alone, the worse ear alone
measurements were not included. This
decision was further influenced by time
constraints with this subject.
Test procedure and materials
The subject used two identical Cochlear
SPECTRA-22 processors, each with its own
ear level microphone, running with indepen­
dent SPEAK strategies (Skinner et al.,1994,
Whitford et aI., 1995) and programmed with
16 and 10 bands on the left and right sides
respectively. The subject had been using the
two processors continually for more than six
months before commencing this study. Both
i
Ii
:1
~,
ii'
iJ
24
monaural (left side alone) and binaural tests
were repeated for two spatially separated
signal and noise configurations: (a) speech
45 degrees to the left and noise 45 degrees to
the right and (b) speech 45 degrees to the
right and noise 45 degrees to the left. These
two configurations will be referred to as L+
and L- conditions respectively, as indicated
in Figure I.
L+ refers to the fact that when the speech
is on the left side (which is the side that
performs better monaurally) and noise on the
right, the left ear has a better input SIN due
to microphone characteristics and head­
shadow effects. Similarly, when the speech is
on the right and noise is on the left, the left
side has a disadvantage due to these effects.
To estimate the actual input SIN at each ear,
spatial head-mounted microphone measure­
ments were used. These were obtained from
an earlier experiment by the authors using a
KEMAR maniquin (Figure 3, p. 250 I, van
Hoesel and Clark, 1994). The measured
characteristics showed, for a signal at 45
degrees (ipsilateral), about 2 dB gain
compared to a 0 degrees direct ahead refer­
ence. At -45 degrees (contralateral) the
characteristics showed about -4 dB gain.
Both speech and noise levels were held
constant throughout the experiment at a SIN
of 5 dB when measured at the centre of the
head (in absentia). For the L- condition this
resulted in a SIN of about 3 dB at the left ear
and about 9 dB at the right ear. Similarly, for
the L+ condition the same SIN ratios resulted
but in the reversed ears. Although the
subject's head was not restrained, he was
always instructed to look directly ahead.
Speech material consisted of open set
CUNY sentences (Boothroyd et aI., 1985)
which were presented at 70 dB SPL (A­
weighted). Multi-talker babble noise
(Auditec, St. Louis, catalogue No. CI46-MT
multitalker) was presented at 65 dB. Both
loudspeakers were at a distance of about 1.4
m from the subject in a mildly echoic test
booth which had an average reverberation
time (T60) of about 0.5 sec. A more detailed
SPEECH RESULTS WITH A COCHLEAR IMPLANT SUBJECT FOR SPATIALLY SEPARATED SIGNAL AND NOISE
/,,>
("''>,
~/S
~>
ON
NV
-0
-0
(a) left L+
(b) left L­
~I
I
N~>
/",/
,"
'~/s
-0­
(e) bin L+
/'
s<:)
ON
-(j­
~
(d) bin L­
FIGURE lA-lD
Test configurations to evaluate better monaural side (left) and binaural listening with 10 em spatially separated
signal (5) and nOise (N) sources (a) left monaural test configuration for Signal on the left, nOise on the right (L+). (b)
left monaural test configuratIOn for signal on the right, nOise on the left (L-) (c) binaural test condition for signal on
the left, no',se on the right (L+) and (d) binaural test conditIOns for signal on the right, nOise on the left (L-)
description of the test booth characteristics is
available in Figure 2(a) and 2(c) in van
Hoesel and Clark, (1995).
Each of the four listening conditions shown
in Figure I. was tested once per session with
a single list of 12 sentences, with a few
sentences for training and acclimatisation
(from additional lists) before each list. Five
data collection sessions were conducted
across five consecutive weeks. Two training
sessions, which were in the same format as
the test sessions, preceded the data collection.
The order of test conditions within each
session were ordered in a 'Latin rectangle'
design to attempt to balance learning and
fatigue effects across the conditions.
For the binaural listening conditions, the
subject was allowed to adjust the sensitivity
control on each processor until the two sides
were perceived as equally loud and the
overall level was comparable to that
preferred by the subject in day to day conver­
sational use. The balance between the two
sides was then verified by presenting babble
noise to either the left, the right or to both
loudspeakers at equal intensity and asking
the subject to identify the source of the sound
as left, right or both speakers. During this
process the subject was requested to look
directly ahead (0 degrees direction). The
processors were adjusted until the subject
was repeatedly able to identify the source(s)
correctly. For monaural listening only the
left processor was switched on and the
subject was allowed to alter its sensitivity for
a comparable loudness to the binaural case.
Note that this means that in the binaural case
the processors were operating at a slightly
lower input sensitivity than in the monaural
case due to binaural loudness summation
effects (van Hoese! & Clark, 1997).
25
-
RICHARD JM van HOESEL AND G.M. CLARK
% Correct
*
l+
l·
( Signal and Noise spatial configuration)
FIGURE 2
Shows averaged res~lts from five test lists of CUNY
sentences, presented at 70 dB, in the presence of
multitalker babble noise at 65 dB. Two test configura­
tions are illustrated signal on the left with nOise on
the right (L+) and signal on the right with noise on the
left (L-) The asterisk indicates a statistically significant
difference between monaural and binaural results.
RESULTS AND DISCUSSION
The results of the L+ (speech towards the left
ear) and L- (speech towards the right ear)
configurations are shown in Figure 2.
The binaural condition shows clear advan­
tages over the monuaral (left only) condition
for the L- configuration. One-way analysis of
variance of the L- results showed a statistically
significant difference between binaural and left
only conditions at p = 0.004. For the L+ tests,
no significant difference was observed between
binaural and monaural results. This is also
important since it shows that, even though the
two sides had different insertion depths, the
subject was not significantly disadvantaged by
the binaural configuration compared to the
better monaural side alone.
Since the subject has previously shown
poor inter-aural time delay discrimination, it
is assumed that intelligibility advantages due
to binaural masking level differences are
minimal, if at all present. In that case, in
modeling how the subject makes use of the
signals at the two ears, we are primarily
concerned with how the signals are weighted
before summation. At one extreme, the two
:t
:I,
~i
~li
:r'1'
:1:
~!
II
I,\1
I
,I
:1
it
;1
\J.
26
ears can be weighted similarly, while at the
other extreme, only one side is attended
while the other side is ignored. For equal
weighting of the two ears, however, the total
internal SIN remains the same for the L+ and
L- configurations. This is the case even
when the performance difference between
the two ears is taken into account by using a
fixed additive internal noise source on each
side. The identical internal binaural SIN ratio
that results for equal weighting is in obvious
contrast to the results obtained, which clearly
show a large difference between L+ and L­
binaural listening conditions. Although it is
possible that the performance difference
between the two sides cannot be modeled by
simple additive internal noise sources, or that
binaural unmasking processes are involved
which do not require interaural time delays
to be preserved, it seems more likely that the
subject was instead applying unequal
weights to the two sides. The most likely
weighting being that which optimizes the
overall internal SIN. This amounts to select­
ing the side which has the higher SIN (allow­
ing for the performance difference on the
two sides) and ignoring the contralateral
side. Although the worse ear measuremcnts
were not made in the present study, some
support that the subject always attended the
ear with the better SIN ratio comes from a
comparison of these binaural results with the
earlier SIT sentence tests in quiet (see patient
details section). For those tests, the subject
showed 60% and 20% on left and right sides
respectively. Given that in the present study
the left ear under the L+ condition receives
the same SIN ratio (about 9 dB) as the right
ear under L-, it is interesting that binaural
results for L+ and L- similarly show results
of 66% and 16% respectively.
Irrespective of the exact mechanism, it is
clear that binaural advantages can be
obtained for this binaural implant user when
signal and noise are spatially separated.
Simple rotation of the head over the range
-90 to +90 degrees will allow the subject to
optimise his SIN under a wide range of
conditions (as do normal hearing subjects
SPEECH RESULTS WITH A COCHLEAR IMPLANT SUBJECT FOR SPATIALLY SEPARATED SIGNAL AND NOISE
---- -----------------------
under adverse noise conditions). This is
possible for monaural implant users only
when the signal is on the same side as the
monaural implant microphone.
The results obtained are encouraging and
lead us to believe that further exploration
with a small group of bilaterally implanted
subjects may be warranted. Further studies
along these lines are being planned and are
expected to include performance evaluation
of both monaural sides as well as binaural
listening for spatially separated signal and
noise. By measuring monaural performance
on both sides for speech on the left with noise
on the right, as well as for the reversed
configuration, the better performing side is
then known for each test configuration. As an
example, although the left side may be the
better performing side for a subject in quiet,
when the noise is on the left, either side may
show better performance depending on the
performance difference between the two
sides. If selective attention to the better
performing monaural side for each spatial
configuration is indeed the predominant
mode of binaural listening in noise, then the
binaural result should always be similar to the
better monaural side for that configuration. If
binaural performance is less than the better
monaural side, then the monaural best side
can not be attended. If binaural performance
is actually better than the better effective SIN,
additional binaural processing may be further
enhancing the signal estimate in background
noise. However, the last case would probably
only be possible for subjects who demon­
strate substantially better inter-aural time
delay discrimination than has been measured
so far with our bilateral implantees.
CONCLUSIONS
A significant improvement in speech intelli­
gibility was observed for binaural listening
compared to monaural listening with the
better ear alone when the noise was on the
same side as the better ear and the signal was
on the contralateral side. When signal and
noise locations were reversed the binaural
performance was similar to the better monau­
ral side alone. The results show that binaural
implantation can offer improved performance
for speech in noise when signal and noise are
spatially distinct and positioned such that the
head-orientation can be used as a means of
improving the SIN at one of the two ears.
ACKNOWLEDG M ENTS
The authors wish to acknowledge the efforts
of the research subject who participated in
this study, Dr. Bob Cowan at the CRC for
Cochlear Implant, and Speech and Hearing
Research, and Dr. Teresa Ching. Financial
assistance for this work was via the Coopera­
tive Research Centre for Cochlear Implant,
Speech and Hearing Research.
REFERENCES
Blarney P. J., Dowell R. C.. Clark a. M. and Seligman
P. M., 1987. Acoustic parameters measured by a
formant-estimating speech processor for a multiple­
channel cochlear implant. J.Acoust.Soc.Am. 82,
48-57.
Boothrooyd A" Hanin L., Hnath T., 1985, A sentence
test of spcech perception: reliability, set equiva­
lence. and short term learning. Internal report RCI
10, Speech and hearing Sciences Research Center.
City University of New York.
Hirsh I. J., 1950, The relation between localization and
intelligibility, J Acoust.SocAm. 22, 196-200.
Licklider J.CR., 1948, The influence of interaural phase
upon the masking of speeCh by whitc noise,
J.AcousI.SocAm.20. 150-159
Marsh M.A., Xu J., Xu S-A., Blarney P.J., Abonyi J.,
Clark a.M., 1993, "Radiological evaluation of multi
channel intra cochlear implant insertion depth", Am.
J. 0101. 14 (4), 386-391.
McKay C, McDermott H, Vandali A, Clark a, 1992, A
comparison of speech perception of cochlear
implantees using thc Spectral Maxima Sound
Processor (SMSP) and the MSP (Multipeak) proces­
sor, Acta Otolaryngol. (SlOckh), 112. 752-761.
Skinner, M.W., Clark a.M., Whitford, L.A., Seligman,
P., Staller, S.J., Shipp, D.B., Shallop, J.K., Evering­
ham. C, Menapace, CM., Arndt, P.L., Antogcnelli,
T., Brimacombe, J.A., Beiter, AL.. 1994, Evalua­
tion of a new spectral peak (SPEAK) coding strat­
egy for the Nucleus multi-electrode cochlear
implant", Ear Hear. 12, 3-22.
van Hoesel R.J.M .• 1998, Bilateral electrical stimulation
with multi-channel cochlear implants, Ph.D. Thesis,
Department of Otolaryngology, The University of
Melbourne.
27
RICHARD J.M van HOESEL AND G.M. CLARK
van Hoesel, R. J. M., Clark, G. Moo 1997, Psychophysi­
cal studies with two binaural cochlear implants
subjects,l.Acoust.Soc.Am. 102,504-518.
van Hoesel, R. J. M., Clark, G. M., 1995, Evaluation of
a portable two-microphone adaptive beam forming
speech processor with cochlear implant patients.
l.!\coust.Soc.!\m. 97,2498-2503.
van Hoesel, R. J. M., Tong, Y. C, Hollow. R. D., Clark,
G. Moo 1993, "Psychophysical and speech percep­
tion studies: A case report on a binaural cochlear
implant subject", 1.Acoust.Soc.Am. 94, 3178-3189.
28
Whitford L A, Seligm~n P M, Everingham C E.
Antognelli T, Skok M C, Hollow R D. Plant K L,
Gerin E S, Staller S J, McDermott H J, Gibson W
R, Clark G M, 1995, "Ev~luation of the Nucleaus
Spectr~ 22 processor and new speech processing
strategy (SPEAK) in postlinguistically deafened
adults", Acta. OWlarynl?ol (Stockh). 115,629-637.
Zurek, P.M., 1983, A predictive model for binaural
advantages in speech intelligibility. l.!\cOUJ/. Soc.
Am. 71, S87.