the nal-nl2 prescription method for hearing...

Harvey Dillon NAL-NL2

Sing Soc Audiol Professionals 1

The NAL-NL2 prescription method for hearing aids

Harvey Dillon, Gitte Keidser, Teresa Ching,

Matt Flax, Scott Brewer

HEARing CRC

National Acoustic Laboratories, Australia

Audiology Online

June 2012

Talk structure

Temporal resolution

Frequency resolution

Intelligibility in noise

OAEsDead

regionsThresholds

NL2 formula

NAL-NL2 derivationIntelligibility

Age

Gender

Experience

Binaural

Language

Speed

NAL NL1

(?)

Amplification rationales

Children

Infants

RECD

nHL eHL

Gain preferences

Outcomes

Loudness NAL-NL1 deficiencies

Severe profound

CR

The Fitting Process



Measure hearing thresholds (dB HL)

Enter into manufacturer software

Adult Child

Measure hearing thresholds (dB HL or dB SPL)

Measure individual RECD (or estimate from age)

Enter into manufacturer software

Verify with real ear measurement

(hearing aid auto adjusted to approximate prescription)

Adjust amplification to better match prescription

(hearing aid auto adjusted to approximate prescription)

Adjust hearing aid in coupler via computer to better match prescribed coupler gain

Prescription goals

Prescribe hearing aids to:Prescribe hearing aids to:

Make speech intelligibleMake speech intelligible Make loudness comfortableMake loudness comfortable

Prescription affected by other things Prescription affected by other things –– localization, localization, –– tonal quality, tonal quality, –– detection of environmental sounds, detection of environmental sounds, –– naturalness. naturalness.



Hearing aids amplify , so ….Hearing aids amplify , so ….

How much amplification?How much amplification?

Gain

50 dB SPL

65 dB SPL

Frequency

80 dB SPL

The lolly shop

Deriving NAL-NL2



The rationale for NAL proceudresThe rationale for NAL proceudres

Maximize calculated speech intelligibility , Maximize calculated speech intelligibility , butbut

Keep total loudness less than or equal to normalKeep total loudness less than or equal to normalKeep total loudness less than or equal to normalKeep total loudness less than or equal to normal

NAL-NL1 (1999) empirical studies

psychoacoustic studies

speech intelligibility models

NAL-NL2

Compare

Deriving optimal gains Deriving optimal gains -- step 1step 1

Speech spectrum & level

Gain-frequency

Loudness model

Normal loudness

Compareresponse

Amplified speech spectrum

Loudness model

Loudness (hearing impaired)

Audiogram

Intelligibility model

Intelligibility achieved

Inverted hearing loss profiles used

Rejection criterion :-30<= G <=60 , where G is the slopesum(H(f))/3 <=100 , where f is in the set {0.5, 1, 2} kHz

The audiograms



The audiograms, continued

Deriving optimal gains Deriving optimal gains -- step 1step 1

Audiogram 1 Speech level 1 Optimal gain frequency response




200 audiograms x 6 speech levels 1200 gain–frequency responses, each at 20 frequencies from 125 Hz to 10 kHz

Overall prescription approachOverall prescription approach

Theoretical predictions

Psychoacoustics

Assumptions, rationale

Final formula

Adjust

Empirical observations

Speech science

Compare



MultiMulti--dimensional equation:dimensional equation:

A neural networkA neural network

H250 H500 H1000 H2000 H8k SPL

G250 G500 G1000 G2000 G8k

The two key ingredientsThe two key ingredients1.1. A loudness modelA loudness model2.2. An intelligibility modelAn intelligibility model

Speech spectrum & level

Loudness model

Normal loudness

CompareGain-frequency response

Amplified speech spectrum

Loudness model

Loudness (hearing impaired)

Audiogram

Intelligibility model

Intelligibility achieved

Calculating loudnessCalculating loudness

Loudness model of Moore and Glasberg (2004)Loudness model of Moore and Glasberg (2004)Allowance for hearing loss

Sumacrossbands

Totalloudness

Calculateloudnessper band

Loudness per band

Filtering intoauditory bands

Excitationlevel

External &middle ear

Free field speech level

Input to cochlea



Speech intelligibility

Freq

1/3

octa

ve S

PL

30

Importance: 0.001 0.002 0.003 0.002

x x x x

Freq

Audibility: 5 16 17 0... ... ... ...

= 0.30

= = = =

0.005 0.032 0.051 0... ... ... ...

Speech Intelligibility IndexSpeech Intelligibility Index

SII =SII = ∑ A∑ Aii IIii

Sum

SII SII ∑ A∑ Ai i IIii

But intelligibility gets worse if we make speech too loud!

Audibility Importance



Speech intelligibility also depends on … Speech intelligibility also depends on … Level distortion Level distortion

Normal hearing people perform poorer at Normal hearing people perform poorer at high speech levelshigh speech levels

tor

Speech level (dB SPL)Leve

l dis

tort

ion

fact

0

1

730 140

SII = SII = ∑ A∑ Ai i IIi i LLii

Level distortion factor

The transfer function

40

60

80

100

nt

Co

rrec

t

Nonsense syllables

Sentences

0

20

40

0 0.2 0.4 0.6 0.8 1

Speech Intelligibility Index (SII)

Per

cen Nonsense syllables



1400 - 5600 Hz

corr

ect

60

80

100

MF MS SF SS

Observed and Predicted performance

Sensation level (dB)

Per

cent

0

20

40

12 24 36 12 24 36 12 24 36 12 24 36

Ching, Dillon & Byrne, 1998

SubjectsSubjects

20 adults with 20 adults with normal hearingnormal hearing

55 adults with 55 adults with sensorineural sensorineural

Audiogram

el (

dB

HL)

-20

0

20

hearing losshearing loss–– mild to mild to

profoundprofound–– Experienced Experienced

hearing aid hearing aid usersusers

Frequency (Hz)

Hea

ring

thre

shol

d le

ve

250 500 1000 2000 4000 8000

40

60

80

100

120

Speech perceptionSpeech perception Stimuli: Filtered speechStimuli: Filtered speech

–– CUNY sentencesCUNY sentences–– VCV syllablesVCV syllables

Shaping:Shaping:–– POGO prescriptionPOGO prescription

20

40

60

80

evel (

dB

SP

L)

HP7 HP14 HP28 LP7LP14 LP28 LP56

p pp p Conditions:Conditions:

–– Quiet at high and low Quiet at high and low sensation levelssensation levels

–– Babble Noise Babble Noise Headphones: Headphones:

Sennheiser HD25Sennheiser HD25

0

20

250 500 1000 2000 4000 8000

Frequency (Hz)

Le



VCV

ect

0 6

0.8

1.0

Audibility and Speech intelligibility Audibility and Speech intelligibility –– H.I. H.I.

Calculated SII

Pro

po

rtio

n c

orr

e

LP7: QLP7: NLP14: QLP14: NLP28: QLP28: NLP56: QLP56: NHP28: QHP28: NHP14: QHP14: NHP7: QHP7: N0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

Deficit = SDeficit = Sansiiansii -- SIISIIeffeff

40

60

80

100

nt

Co

rre

ct

Deficit = 0.6 - 0.4

= 0.2

0

20

40

0 0.2 0.4 0.6 0.8 1

Speech Intelligibility Index (SII)

Per

cen

SIIeff SIIansi

VCV deficit vs CUNY deficitVCV deficit vs CUNY deficit

0.05

0.10

0.15

0.20

0.25

SIIe

ff

R=0.77

-0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 0.25 0.30

VCV SIIansi-SIIeff

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

CU

NY

SIIa

nsi-S



Intelligibility and audibilityIntelligibility and audibility

1

30Sensation level (dB)

m

p

BKB, VCV and CUNY

BKB VCV CUNY

Q

Optimizer results: 3 data sets

Q&N



Desensitisation for hearing lossDesensitisation for hearing loss

0 40.50.60.70.80.9

1

ve a

ud

ibili

ty 020406080

00.10.20.30.4

0 20 40 60

Sensation level (dB)

Effe

cti 80

100120

0

0.2

0.4

0.6

0.8

1

0 20 40 60 80 100 120

Hearing threshold (dB HL)

m p

aram

eter

Psychoacoustics

Why measure only pure tone thresholds?Why measure only pure tone thresholds?

Other measurementsOther measurements

Outer hair cell function Outer hair cell function –– clickclick--evoked otoacoustic emissionsevoked otoacoustic emissions

Frequency resolution Frequency resolution –– psychophysical tuning curvespsychophysical tuning curves–– cochlear dead regions cochlear dead regions –– TEN testTEN test

Cognitive abilityCognitive ability AgeAge



Healthy PTC Healthy PTC –– no dead regionno dead region

Psychophysical tuning curve

SP

L)

90

100

110 A29

Masker

Masker Frequency (Hz)

Mas

ker

Leve

l (dB

S

500 1000 1500 200050

60

70

80

90

Signal

Poor PTC: Dead region at 4 kHzPoor PTC: Dead region at 4 kHzPsychophysical tuning curve

SP

L)

110

120

Masker Frequency (Hz)

Mas

ker

Leve

l (dB

1000 2000 3000 4000 500080

90

100

Dead regionsDead regions

NAL-NL1 only allows for hearing loss desensitization on average

RIP



OffOff--frequency listening: TEN testfrequency listening: TEN test

e vi

brat

ion

Frequency or position

Basi

lar

mem

bran

e

Based on Moore (2004)

Threshold Equalizing Noise (TEN)

TEN and PTC (non) agreementTEN and PTC (non) agreement

2 kHz2 kHz TEN: AliveTEN: Alive TEN: DeadTEN: Dead

PTC: PTC: 6060 11

TEN

uncertain

1Tip in placeTip in place

PTC:PTC:Tip shiftedTip shifted

44 33

PTC uncertain 1

2

2

1

1

Psychoacoustic correlations Psychoacoustic correlations –– 4 kHz4 kHzMatrix Plot (Prof and Psy 11 March 09.sta 659v*75c)

HL

PTC Q10

OAE strength

TEN elevation

Cognition

Age



Psychoacoustic correlations Psychoacoustic correlations –– 2 kHz2 kHzMatrix Plot (Prof and Psy 11 March 09.sta 659v*75c)

HL2k

PTC2k_Q10

C2kC2k

El2k

Cognition

AgeLim

Can we better predict Can we better predict intelligibility if we use intelligibility if we use psychoacoustic results?psychoacoustic results?psychoacoustic results?psychoacoustic results?

Deficit (VCV & CUNY), HL, Q10, OAE, TEN, Cog, & Age: 5600 Hz low passes-SIIeff (VCV Avg(QH QL N) LP5600)

SIIdes-SIIeff (CUNY Avg(QH QL N) LP5600)

L56HL

Q10 LP56

COAE L56

EL LP56

Cogni tion

AgeLim



Yes, a little Yes, a little –– speech speech deficit increases as deficit increases as frequency selectivity gets frequency selectivity gets broaderbroader

But not once we But not once we fully build HL into fully build HL into the SII prediction the SII prediction

Correlations

HL 500 PTC5_Q10

OAE 500

TEN 500

Cognit Age

HL 500 -- -0.70 0.04 0.56 -0.38 0.26PTC5_Q10 -0.70 -- -0.06 -0.48 0.33 -0.17OAE 500 0.04 -0.06 -- -0.10 -0.05 -0.05TEN 500 0.56 -0.48 -0.10 -- -0.35 0.04Cognition -0.38 0.33 -0.05 -0.35 -- -0.45A 0 26 0 17 0 05 0 04 0 45

500 HzHL 1k PTC1k

_Q10OAE 1k

TEN 1k

Cognit Age

HL 1k -- -0.63 -0.36 0.46 -0.38 0.30PTC1k_Q10 -0.63 -- 0.30 -0.10 0.23 -0.17OAE 1k -0.36 0.30 -- -0.18 0.20 -0.20TEN 1k 0.46 -0.10 -0.18 -- -0.13 0.10Cognition -0.38 0.23 0.20 -0.13 -- -0.45Age 0 30 -0 17 -0 20 0 10 -0 45 --

1 kHz

Age 0.26 -0.17 -0.05 0.04 -0.45 -- Age 0.30 0.17 0.20 0.10 0.45

HL 4k PTC4k_Q10

OAE 4k

TEN 4k

Cognit Age

HL 4k -- -0.84 -0.60 0.58 -0.43 0.41PTC4k_Q10 -0.84 -- 0.50 -0.49 0.33 -0.23OAE 4k -0.60 0.50 -- -0.31 0.44 -0.40TEN 4k 0.58 -0.49 -0.31 -- -0.45 0.38Cognition -0.43 0.33 0.44 -0.45 -- -0.45Age 0.41 -0.23 -0.40 0.38 -0.45 --

4 kHz2 kHz

HL 2k PTC2k_Q10

OAE 2k

TEN 2k

Cognit Age

HL 2k -- -0.79 -0.65 0.53 -0.40 0.35PTC2k_Q10 -0.79 -- 0.49 -0.47 0.36 -0.25OAE 2k -0.65 0.49 -- -0.36 0.35 -0.36TEN 2k 0.53 -0.47 -0.36 -- -0.38 0.15Cognition -0.40 0.36 0.35 -0.38 -- -0.42Age 0.35 -0.25 -0.36 0.15 -0.42 --

CorrelationsCorrelations

PTCAge

HL

Cognit

OAE

TEN



Multiple regressionMultiple regression

PTC

including HL causes:including HL causes:

correlations between age and PTC / OAE / TEN to disappearcorrelations between age and PTC / OAE / TEN to disappear

correlations between cognition and PTC / OAE / TEN to disappearcorrelations between cognition and PTC / OAE / TEN to disappear

Age OAE

TEN

Cognition

HL

Likely intermediate effectsLikely intermediate effects

? PTC

OAE

Cognition

Mechanical

Age

Cardio-vascular

IHC

Noise

OAE

TEN

HL

OHC

Stria

Why are hearing thresholds so useful?Why are hearing thresholds so useful?

H i Speech

Frequency selectivity

Temporalresolution

Hearing thresholds

Speech Perceptionproficiency

Central auditory processing

OtherAge

Cognitive ability



Factors affecting prescription

National Acoustic Laboratories, Sydney, Australia Keidser and Dillon

Gain; 187 adults, medium input level


Source: Keidser et al., 2008




Adjustments to prescription to allow for experience

68

1012

men

t NewExperienced

-8-6-4-20246

0 20 40 60 80 100 120


Gai

n a

dju

stm


Adjustments to prescription to allow for experience

68

1012

men

t

New

Experienced

Experienced

-8-6-4-20246

0 20 40 60 80 100 120


Gai

n a

dju

stm Experienced -

New


Gain preference over time

N = 11

Source: Keidser et al. (submitted)




0.5

1

1.5

atio

n f

rom

ef

erre

d a

t 65

d

B

Smeds et al. 2006

Zakis et al. 2007

Gain; adults, low and high input levels

-2

-1.5

-1

-0.5

0

Pre

ferr

ed g

ain

dev

iN

AL

-NL

1 re

gai

n p

r ed

B S

PL

in

d

50 80

Input level in dB SPL

Output level

Adults, NL2Children, NL2

NAL-NL1

Desired gain

Gain and compression; adults vs children

Input level

Adults, NL2

Age dependent gain


Adults – congenital or acquired?

-4

-2

0

evia

tio

n

(dB

)

-14

-12

-10

-8

-6

Pre

ferr

ed g

ain

de

fro

m N

AL

-RP

LFA HFA

Congenital(N=15)Acquired(N=28)



The maximum C

Compression

Limiting compression for severe/profound hearing loss(Fast compression)

Source: Keidser et al., 2007

NAL-NL2 prescribes

creating sound valueTM

0

50

100

H 2

4

6

log2�

�

�� f

125

�

�

��

1.0

1.5

2.0

2.5

3.0

Ca

NAL NL2 prescribes different compression for slow and fast compressors for those with high degrees of hearing loss


Compression ratio preferences: severe and profound hearing loss

0.8

0.9

1.0

1.1

LF b

and

1:1

45 50 55 60 65 70 75 80 85 90

Average HTL in LF band (dB HL)

0.2

0.3

0.4

0.5

0.6

0.7

Ave

rage

1/C

R in

L

Source: Keidser, Dillon, Dyrlund, Carter, and Hartley (2007)

1.8:1

3:1

Gain preference by aid configuration (N = 187)

Bilateral loudness correction

Bilateral loudness correction too large – supported by newer

data (e.g. Whilby, 2006; Epstein & Florentine, 2009)



New bilateral loudness correction

3

4

5

6

7

riat

ion

(d

B)

0

1

2

3

0 20 40 60 80 100 120

Input level (dB)

Gai

n v

a


Effect of language

Gain at each frequency depends on importance of each frequency

Low frequencies more important in tonal Low frequencies more important in tonal languages

Two versions of NAL-NL2Tonal languages

Non-tonal languages

15

20

25

30

35

40

tion gain in dB

NAL‐NL1

Male, exp, non‐tonal

Male, exp, tonal

Tonal versus non-tonal language

0

5

10

15

100 1000 10000

Insert

Frequency in Hz



RECD in infants (own mold; HA2)

15

20

25

30

35

CD

_4

k

0.60.8

2.04.0

6.08.0

20.040.0

60.080.0

200.0400.0

Age (months)

-5

0

5

10

15

RE

C

• New features in NAL-NL2– Different gain-frequency response shape and

higher compression ratios

– Different compression ratios for fast and slow compressors (severe/profound hearing loss)

Summary

– Gender dependent gain

– Age dependent gain

– Gain adaptation for new hearing aid users

– Language dependent gain (tonal vs non-tonal)

Examples of prescriptions



Example audiogram: moderate sloping

303540

n (

dB

)

250125 500 1k 2k 4k 8k

0

20

Frequency (Hz)

HL)

50 dB

05

10152025

100 1000 10000

Frequency (Hz)

Inse

rtio

n G

ain

40

60

80

100

120

Hea

ring

thre

shol

d (d

B

65 dB

80 dB

303540

n (

dB

)

250125 500 1k 2k 4k 8k

0

20

Frequency (Hz)

HL) 50 dB

Example audiogram: flat 60

05

10152025

100 1000 10000

Frequency (Hz)

Inse

rtio

n G

ain

40

60

80

100

120

Hea

ring

thre

shol

d (d

B

65 dB

80 dB

303540

n (

dB

)

250125 500 1k 2k 4k 8k

0

20

Frequency (Hz)

HL)

50 dB

Example audiogram: steeply sloping

05

10152025

100 1000 10000

Frequency (Hz)

Inse

rtio

n G

ain

40

60

80

100

120

Hea

ring

thre

shol

d (d

B

65 dB

80 dB



25

30

35

n (

dB

)

250125 500 1k 2k 4k 8k

0

20

Frequency (Hz)

HL)

50 dB

Example audiogram: extreme ski-slope

0

5

10

15

20

100 1000 10000

Frequency (Hz)

Inse

rtio

n G

ain

40

60

80

100

120

Hea

ring

thre

shol

d (d

B

65 dB

80 dB

25

30

35

n (

dB

)

250125 500 1k 2k 4k 8k

0

20

Frequency (Hz)

HL) 50 dB

Example audiogram: reverse sloping

0

5

10

15

20

100 1000 10000

Frequency (Hz)

Inse

rtio

n G

ain

40

60

80

100

120

Hea

ring

thre

shol

d (d

B

65 dB

80 dB

Figure 5. Specific loudness for A-1 through A-4 based on a 65 dB international long-term average speech spectrum input in quiet. Overall loudness is shown to the right of the prescriptive method in the legends (NH designates normal hearing).

Johnson & Dillon, JAAA, 2011



Figure 5. Specific loudness for A-1 through A-4 based on a 65 dB international long-term average speech spectrum input in quiet. Overall loudness is shown to the right of the prescriptive method in the legends (NH designates normal hearing).


Figure 7. Overall loudness of each prescriptive method averaged across the five sensorineural hearing losses (A-1 through A-5) based on a 65 dB international long-term average speech spectrum input in quiet.


Figure 8. Average Speech Intelligibility Index (SII) value for speech in quiet across the five sensorineural hearing losses for each prescriptive method using both the ANSI S3.5-1997 and the National Acoustic Laboratories SII methods. Also shown is the SII transformed value into a predicted speech recognition score (% correct) for the Connected Speech Test (Cox et al, 1987) using the transfer function of Humes (2002).




We measure several times,

T Ching, NAL, CRC HEAR

NAL & DSL groups have

equal language outcomes

Relationships between variables in NAL-NL2

REDD

N

BWC

WBCT

REUG

Bi-uni Gender

Age

ExperienceRECD

MLE

Aid type

I/O

Vent Tube

UCT

CR

Comp speed

Language DepthRECD

AC

BC'

ABG

BC

REIG REAG

RESR

CG

MLE

SSPL2cc

Limitingtype

SSPLES

ESCD

Blue = User i/p

Grey = internal variable

Red = effect of saturation

Dash-dot = alternatives

Green = stored data

AC'

ESG

Transducer

Taking the pressure off prescription ……

The trainable hearing aid



Trainable Aid

Aid user adjusts settings...

Dillon et al (2006)

Zakis et al (2007)

Trainable Aid

Process repeats for other sounds

Trainable Aid

After training...

...preferred settings are automatically applied



Training gain, CR, CT.

15

20

25

30

n (

dB

) CRGain

0

5

10

30 40 50 60 70 80 90

Input level (dB SPL)

Gai

n

CT

“A challenge for the profession is to devise fitting procedures that are scientifically defensible and the challenge for thethe challenge for the individual audiologist is to choose the best procedures from whatever are available”

Denis Byrne, 1998

Acknowledgements

creating sound valueTM

This research was financially supported by the HEARing CRC established and supported under the Australian Government’s Cooperative Research Centres Program

www.hearingcrc.org

www.nal.gov.au

the nal-nl2 prescription method for hearing...

Documents