Upload
madyson-greenwell
View
229
Download
0
Tags:
Embed Size (px)
Citation preview
Comparative performance of Comparative performance of an an adaptive directional microphoneadaptive directional microphone and and a a multi-channel noise reductionmulti-channel noise reduction systemsystem, , in a digital hearing instrumentin a digital hearing instrument
Kevin CP YuenKevin CP YuenInstitute of Human Communicative Research & Institute of Human Communicative Research & Division of Otolaryngology, Department of Surgery, Division of Otolaryngology, Department of Surgery, The Chinese University of Hong KongThe Chinese University of Hong Kong
Anna CS KamAnna CS KamPhonak Hong Kong Hearing CentrePhonak Hong Kong Hearing Centre
Polly SH LauPolly SH LauDivision of Speech and Hearing Sciences, Division of Speech and Hearing Sciences, The University of Hong KongThe University of Hong Kong 8 Oct 20058 Oct 2005
2
ObjectivesObjectives
Compare the speech recognition in noise perfCompare the speech recognition in noise performance amongormance among Multi-channel noise reduction system (medium) (Multi-channel noise reduction system (medium) (FNCFNC),), Adaptive directional microphone (Adaptive directional microphone (DAZDAZ), &), & FNC + DAZFNC + DAZ
Compare the objective and subjective performCompare the objective and subjective performance betweenance between NAL-NL1NAL-NL1 in Perseo, & in Perseo, & Digital Perception Processing (Digital Perception Processing (DPPDPP) in Perseo) in Perseo
3
Multi-channel noise reduction system -Multi-channel noise reduction system -Fine noise canceller, FNC (Medium)Fine noise canceller, FNC (Medium)
Courtesy of Phonak
Frequency-Frequency-dependent gain dependent gain reductionreduction
SNR-dependent SNR-dependent gain reductiongain reduction
90 200
400
600
800
1000
3000
5000
7000
Frequency (Hz)
0
2
4
6
8
10
12
Att
entu
atio
n (d
B)
Noise only SNR 0 SNR+5 SNR+10 Signal only
4
Adaptive directional microphone (DAZ)Adaptive directional microphone (DAZ) Polar patterns are Polar patterns are
switched switched adaptively in adaptively in response to the response to the direction of noisedirection of noise
Courtesy of Phonak
5
NAL-NL1 & DPPNAL-NL1 & DPP NAL-NL1NAL-NL1, National Acoustics Laboratory – Non-linear , National Acoustics Laboratory – Non-linear
1 (Dillion, 1999)1 (Dillion, 1999) Provides gain-frequency response that maximizes speech intProvides gain-frequency response that maximizes speech int
elligibility for any input level of speech,elligibility for any input level of speech, while keeping overall loudness at a level no greater than that while keeping overall loudness at a level no greater than that
perceived by a normal hearing person, or at a lower level if it perceived by a normal hearing person, or at a lower level if it provides greater speech intelligibility.provides greater speech intelligibility.
DPPDPP, Digital Perception Processing, Digital Perception Processing Incorporates a psychoacoustic loudness mode of normal and Incorporates a psychoacoustic loudness mode of normal and
impaired cochlea function (Moore et al, 2003)impaired cochlea function (Moore et al, 2003) Calculates excitation patterns and specific loudness patterns Calculates excitation patterns and specific loudness patterns
evoked in a normal and impaired cochlear, from the incominevoked in a normal and impaired cochlear, from the incoming signal, and adjusts gain required to match the loudness pattg signal, and adjusts gain required to match the loudness patterns of impaired cochlea to those of normal cochleaerns of impaired cochlea to those of normal cochlea
Aims at restoring loudness perception to ‘normal’Aims at restoring loudness perception to ‘normal’
6
Research DesignResearch Design
Group 1 (5 subjects)Group 1 (5 subjects)
OWN hearing aidOWN hearing aid → → Perseo DPPPerseo DPP →→ Perseo NAL-NL1Perseo NAL-NL1
1 1 wweekeek 1 week1 week
Group 2 (4 subjects) Group 2 (4 subjects)
OWN hearing aidOWN hearing aid → → Perseo NAL-NL1Perseo NAL-NL1 → → Perseo DPPPerseo DPP
Perseo DPPPerseo DPP == Digital Perception Processing targetDigital Perception Processing targetPerseo NAL-NL1Perseo NAL-NL1 = = NAL-NL1 targetNAL-NL1 target
7
Assessment toolsAssessment tools
DPPDPP vs vs NAL-NL1 NAL-NL1 / / FNC+DAZFNC+DAZ vs vs DAZDAZ vs vs FNCFNC Cantonese Hearing-in-Noise Test (CHINT)Cantonese Hearing-in-Noise Test (CHINT)
(Wong & Soli, 2005)(Wong & Soli, 2005) Chinese translated version of Abbreviated Profile of HeChinese translated version of Abbreviated Profile of He
aring Aid Benefit (APHAB)aring Aid Benefit (APHAB) (Cox & Alexander, 1995)(Cox & Alexander, 1995)
DPPDPP vs vs NAL-NL1NAL-NL1 Sound quality paired comparison judgmentSound quality paired comparison judgment
8
SubjectsSubjects
250 500 1000 2000 4000 8000Frequency, Hz
0
20
40
60
80
100
dB
HL
Nine post-lingually deafened adults (39-79 yrNine post-lingually deafened adults (39-79 yrs)s)
Sensorineural hearing lossSensorineural hearing loss Native Cantonese-speakingNative Cantonese-speaking Hearing aid experience > 6 monthsHearing aid experience > 6 months
Own hearing aidPerseo hearing
aidMultiband noise redu
ction
Directional mic
Compression scheme
Ear
S1 OC L 211
S2 OC R 211
S3 OC R 211
S4 ✔ WDRC R 311
S5 OC L 311
S6 ✔ ✔ WDRC R 211
S7 WDRC L 211
S8 WDRC L 211
S9 ✔ ✔ WDRC R 211
Average unaided air-conduction threshold of the test ear (+/- 1 SD)
9
Speech recognition in noise resultsSpeech recognition in noise results
Separately for Separately for Signal_Front/ Noise_Front & Signal_Front/ Noise_Front & Signal_Front/ Noise_SideSignal_Front/ Noise_Side
RunRunTwo-way repeated measures ANOVA forTwo-way repeated measures ANOVA for Noise suppression (Noise suppression (FNC+DAZFNC+DAZ vs vs DAZDAZ vs vs FNCFNC)) Prescription (Prescription (DPPDPP vs vs NAL-NL1NAL-NL1))
&&Post-hoc Tukey HSD testPost-hoc Tukey HSD test
10
ResultsResults
Noise suppression optionsNoise suppression options
FNC+DAZFNC+DAZ vs vs DAZDAZ vs vs FNCFNC
11
FNC+DAZDAZFNC
S3 S4 S5 S6 S7 S8 S9Subject
0
20
40
60
80
100
Me
an
CH
INT
sc
ore
(%
)
S1S1 S2S2 S3S3 S4S4 S5S5 S6S6 S7S7 S8S8 S9S9
SNRSNR -- -- 00 00 +5+5 +10+10 +5+5 -5-5 +5+5
FNC+DAZFNC+DAZ vs vs DAZDAZ
FNC+DAZFNC+DAZ vs vs FNCFNC
DAZDAZ vs vs FNCFNC
FNC+DAZFNC+DAZ vs vs DAZDAZ vs vs FNCFNC Signal_Front / Noise_FrontSignal_Front / Noise_Front
12
FNC+DAZDAZFNC
S3 S4 S5 S6 S7 S8 S9Subject
0
20
40
60
80
100
Me
an
CH
INT
sc
ore
(%
)
S1S1 S2S2 S3S3 S4S4 S5S5 S6S6 S7S7 S8S8 S9S9
SNRSNR -- -- 00 00 +5+5 +10+10 +5+5 -5-5 +5+5FNC+DAZFNC+DAZ vs vs DAZDAZ DAZDAZ
FNC+DAZFNC+DAZ vs vs FNCFNC FNC+FNC+DAZDAZ
FNC+FNC+DAZDAZ
FNC+FNC+DAZDAZ
FNC+FNC+DAZDAZ
FNC+FNC+DAZDAZ
FNC+FNC+
DAZDAZ
DAZDAZ vs vs FNCFNC DAZDAZ DAZDAZ DAZDAZ DAZDAZ DAZDAZ DAZDAZ DAZDAZ DAZDAZ
FNC+DAZFNC+DAZ vs vs DAZDAZ vs vs FNCFNCSignal_Front / Noise_SideSignal_Front / Noise_Side
13
ResultsResults
PrescriptionsPrescriptions
DPPDPP vs vs NAL-NL1NAL-NL1
14
DPPNAL-NL1
S3 S4 S5 S6 S7 S8 S9Subject
0
20
40
60
80
100M
ea
n C
HIN
T s
co
re (
%)
S1S1 S2S2 S3S3 S4S4 S5S5 S6S6 S7S7 S8S8 S9S9
SNRSNR -- -- 00 00 +5+5 +10+10 +5+5 -5-5 +5+5
DPP DPP vs vs
NAL-NL1NAL-NL1
DPPDPP
DPPDPP vs vs NAL-NL1NAL-NL1 Signal_Front / Noise_FrontSignal_Front / Noise_Front
15
DPPNAL-NL1
S3 S4 S5 S6 S7 S8 S9Subject
0
20
40
60
80
100
Me
an
CH
INT
Sc
ore
(%
)
S1S1 S2S2 S3S3 S4S4 S5S5 S6S6 S7S7 S8S8 S9S9
SNRSNR -- -- 00 00 +5+5 +10+10 +5+5 -5-5 +5+5
DPPDPP vs vs
NAL-NL1NAL-NL1
DPPDPP DPPDPP DPPDPP DPPDPP
DPPDPP vs vs NAL-NL1NAL-NL1 Signal_Front / Noise_SideSignal_Front / Noise_Side
16
APHAB APHAB DPPDPP vs vs NAL-NL1NAL-NL1 APHAB subscale
Sums of negative ranks
(NAL-NL1 > DPP)
Sums of positive ranks
(DPP > NAL-NL1)
p value
Ease of communication
323.0 343.0 0.87
Reverberation 70.5 205.5 0.04
Background noise
1129.5 950.5 0.55
Adversiveness 474.5 155.5 0.01
Wilcoxon Signed-Ranks Test
17
Sound quality paired comparison judgmentSound quality paired comparison judgment DPPDPP vs vs NAL-NL1NAL-NL1
SOFT loudnessSOFT loudness MEDIUM loudnessMEDIUM loudness
so1so1 Bird singingBird singing m1m1 Alarm clockAlarm clock
so2so2 CatCat m2m2 Bird singingBird singing
so3so3 Clock tickingClock ticking m3m3 Church bellChurch bell
so4so4 Distant thunderDistant thunder m4m4 Classical musicClassical music
so5so5 Running waterRunning water m5m5 RainingRaining
so6so6 Male radio broadcastMale radio broadcast m6m6 Toilet flushingToilet flushing
so7so7 Female radio broadcastFemale radio broadcast m7m7 Male radio broadcastMale radio broadcast
so8so8 Female singingFemale singing m8m8 Female radio broadcastFemale radio broadcast
so9so9 Male singingMale singing m9m9 Female singingFemale singing
m10m10 Male singingMale singing
18
Sound quality paired comparison judgmentSound quality paired comparison judgment DPPDPP vs vs NAL-NL1NAL-NL1
Track
Subject So1
So2
So3
So4
So5
So6
So7
So8
So9
M1
M2
M3
M4
M5
M6
M7
M8
M9
M10
S1 - D N D N D N D N D D D D D D N D D N
S2 D N D N D D D D D N D D D N D N N N D
S3 N N N D D D D D D - D D N N D D D N N
S4 D D N D N N D D N - D D D D D N D D N
S5 - - D N N D D D D - N D D N D D D N N
S6 D N D N N N D D N - D D D N N N D D N
S7 N N D N D D N D N - D - N N N D N N D
S8 D N D D N D N N N - D D N N N N D D N
S9 D N D N N N D N N - N N D N N D N N D
19
Sound quality paired comparison judgmentSound quality paired comparison judgment DPPDPP vs vs NAL-NL1NAL-NL1
N D
S1 NN NN NN NN NN NN DD DD DD DD DD DD DD DD DD DD DD DD - 6 12
S2 NN NN NN NN NN NN NN DD DD DD DD DD DD DD DD DD DD DD DD 77 1212
S3 NN NN NN NN NN NN NN DD DD DD DD DD DD DD DD DD DD DD - 7 11
S4 NN NN NN NN NN NN DD DD DD DD DD DD DD DD DD DD DD DD - 6 12
S5 NN NN NN NN NN NN DD DD DD DD DD DD DD DD DD DD - - - 6 10
S6 NN NN NN NN NN NN NN NN NN DD DD DD DD DD DD DD DD DD - 9 9
S7 NN NN NN NN NN NN NN NN NN NN DD DD DD DD DD DD DD - - 10 7
S8 NN NN NN NN NN NN NN NN NN NN DD DD DD DD DD DD DD DD - 10 8
S9 NN NN NN NN NN NN NN NN NN NN NN NN DD DD DD DD DD DD - 12 6
NNDD DPP NALWilcoxon Signed-Ranks Test , p =.26
20
S1 S2 S3
S4 S5 S6
S7 S8 S9
DPP - NAL-NL1 @ 50 dBSPLDPP - NAL-NL1 @ 65 dBSPLDPP - NAL-NL1 @ 80 dBSPL
Frequency, Hz
-15
0
15
30
Dif
fere
nce
, d
B
-15
0
15
30
Dif
fere
nce
, d
B
0
25
00
50
00
75
00
-15
0
15
30
Dif
fere
nce
, d
B
0
25
00
50
00
75
00
0
25
00
50
00
75
00
Real ear SPL difference Real ear SPL difference DPPDPP minus minus NAL-NL1NAL-NL1
DPPDPPDPP
DPP
Diff
eren
ce,
dB
21
ConclusionConclusion
Speech recognition in noiseSpeech recognition in noise With With signalsignal and and noisenoise from the from the frontfront,,
DPPDPP better than better than NAL-NL1NAL-NL1 in 1 subject in 1 subject FNCFNC, , DAZDAZ and and FNC+DAZFNC+DAZ were not different were not different
With With signalsignal from the from the frontfront, , noisenoise from the from the sideside DPPDPP better than better than NAL-NL1NAL-NL1 in 4 subjects in 4 subjects DAZDAZ better than better than FNCFNC in 8 subjects in 8 subjects FNC+DAZFNC+DAZ better than better than FNCFNC in 6 subjects in 6 subjects
Whenever there was a difference, Whenever there was a difference, DPPDPP was better than was better than NAL-NL1NAL-NL1 Adaptive Directional MicAdaptive Directional Mic better than better than Multiband noise reduction Multiband noise reduction
systemsystem, only for , only for signal_front/noise_sidesignal_front/noise_side but not for but not for signal_frontsignal_front/noise_front/noise_front
best speech recognition, measured AI-weighted directional pattern best speech recognition, measured AI-weighted directional pattern for for signal_front/noise_side signal_front/noise_side (Ricketts et al, 2002)(Ricketts et al, 2002)
22
ConclusionConclusion
APHABAPHAB NAL-NL1NAL-NL1 > > DPP DPP (Adversiveness)(Adversiveness) DPPDPP > > NAL-NL1NAL-NL1 (Reverberation) (Reverberation)
Sound quality paired comparison judgmentSound quality paired comparison judgment 5/9 subjects had more preferences towards 5/9 subjects had more preferences towards DPPDPP th th
an NAL-NL1an NAL-NL1 3/9 subjects had more preference towards 3/9 subjects had more preference towards NAL-NL1NAL-NL1
than DPPthan DPP
23
NAL-NL1NAL-NL1EnglishEnglish does not maximize speech does not maximize speech
intelligibility for intelligibility for CantoneseCantonese ? ?
Band importance functionBand importance function English sentence – peak at 2000Hz (Eisenberg et al, 1998)English sentence – peak at 2000Hz (Eisenberg et al, 1998) Cantonese sentence – peak at 800-1600Hz (Chua, 2004)Cantonese sentence – peak at 800-1600Hz (Chua, 2004)
Band importance functionBand importance function -> derive -> derive SIISII -> derive -> derive NAL-NL1NAL-NL1
Chua, 2004
24
Kevin YuenKevin Yuen
Division of Otorhinolaryngology, Division of Otorhinolaryngology,
Department of SurgeryDepartment of Surgery
The Chinese University of Hong KongThe Chinese University of Hong Kong
Email: Email: [email protected]@surgery.cuhk.edu.hk
25
2cc coupler gain response of the three hearing aid 2cc coupler gain response of the three hearing aid settings (front microphone @ 0 degree azimuth)settings (front microphone @ 0 degree azimuth)
200
400
600
800
1000
3000
5000
7000
9000
Frequency (Hz)
-40
-30
-20
-10
0
10
20
30
40G
ain
(dB
)
FNC DAZ FNC+DAZ