EVALUATION OF NEW HEARING AID
TECHNOLOGY

by

Linda M. Thibodeau, Ph.D.
Advanced Hearing Research Center
Callier Center for Communication Disorders
University of Texas at Dallas

ACKNOWLEDGEMENTS

FRYE ELECTRONICS

Wayne Staab, Consultant Frye Electronics

University of Texas at Dallas PhD students

Paul Dybala

Gary Overson

Jack Scott

MURPHY’S LAW

OVERVIEW
Part 2

VERIFICATION ISSUES

PROCEDURES FOR TESTING AIDS WITH:

1) DIGITAL AIDS

2) MULTIPLE CHANNELS

3) ADAPTIVE RELEASE TIME

4) DIRECTIONAL MICROPHONES

5) GROUP DELAY and PHASE

6) REDUCED RESIDUAL CANAL VOLUME (CICs)

 

OVERVIEW
Part 3

INTERPRETING SPECIAL CURVES

SO WHAT?

 

 

HEARING AID
FEATURES

BEFORE WE BEGIN WITH EVALUATION OF THESE OPTIONS, LET’S LOOK AT
….SOME STANDARD VERIFICATION TOOLS….

ANSI S3.22-1996 (revision of ANSI ’87)

Specification of Hearing Aid Characteristics

ANSI S3.42-1992

Testing Hearing Aids with a Broad-Band Noise Signal

 

Latest Option:

ANSI S3.22 2003!!!

Won’t see it on Test Boxes until about 2005!

Couplers

Hearing Aid Settings

Adjust controls to the settings stated by the manufacturer or in ANSI if you want to compare to SPECS!

If doing a quick check, can run at user settings. Be sure to make note of settings for future reference.

ANSI S3.22-1996 (revision of ANSI ’87)
Specification of Hearing Aid Characteristics

1) OSPL90

OSPL90 (Old Term: SSPL90)

Output (Saturation) Sound Pressure Level

Present a 90 dB input level to the HA mic

VC = Full-on

Measure over frequency range

 

1) OSPL90

USEFULNESS?

Allows one to see the maximum output the HA can generate

Gives quick estimate of aid’s power (High gain: peak at 130 dB SPL Plus!, Low gain: peak at 120 dB or less!

Can relate to HA matrix (1st number), i.e., 114/40/15

Compare to directly to UCL when measured with insert earphone and expressed in dB SPL

2) HFA OSPL90-High Frequency Average Output Saturation Sound Pressure Level

3) FOG-Full On Gain

Full-on Gain

60 dB SPL input to HA mic for non AGC aids

50 dB SPL input to HA mic for AGC aids

and high power aids

VC = Full-on

3) FOG-Full On Gain

Usefulness?

Allows an estimate of the power of the aid (HIGH: FOG is 50 to 75 dB; MILD gain is 20 to 40 dB

Middle number of HA matrix (114/40/15)

(Last number of the matrix is the rise in the curve in dB between 500 Hz and the first peak on the graph)

4) HFA FOG- High Frequency Average Full-On Gain Curve

5) SPA FOG-Special Purpose Average

Special Purpose Hearing Instrument Average (SPA)

Used for very high frequency response hearing aids having essentially no gain at 1000 or 1600 Hz

Manufacturer determines the 3 frequencies to use for these to calculate a SP Average

6) RTG-Reference Test Gain

RTGP-Reference Test Gain Position

Reduced VC setting to simulate how aid performs at a level closer to typical speech

Conversational speech of 65 dB SPL has ± 12 dB variation so max level would = 77 dB;

HFA OSPL90 – 17 dB = Target level to adjust VC to achieve with a 60 dB SPL input

RTG-Explanation

We want to turn down volume to a "use" setting….

So that could be the setting that allows the typical input level (65) plus variation (+12) to not exceed OSPL90;

So if we want to find that level…then work backwards…Take OSPL90 minus 65 minus 12

BUT

We use 60 dB input….so we subtract 17 dB from HFA OSPL90 and adjust VC to get that output with average input at 1, 1.6, and 2.5 kHz at 60 dB SPL

7) Frequency Response

Reference Test Gain=HFA OSPL90-17-60 or

=HFA OSPL90-77

Frequency Response Curve-Set VC to RTG

Input = 60 dB SPL

Input = 50 dB Low gain aids and AGC aids

Frequency Range

Determined from Frequency Response Curve

Average of 1000, 1600, and 2500 Hz, and subtract 20 dB.

Draw horizontal line. Where horizontal line crosses response curve at left is F1 (lowest frequency) and on right is F2 (highest frequency)

Frequency Range Calculation

8) Equivalent Input noise

The noise made by the circuit of the hearing aid!

The way it is reported is misleading….

The number represents the equivalent amount of noise to deliver to the HA mic to result in the noise output that equals the circuit noise

EIN=Level in coupler with no input… minus HFA output… minus 60 dB input or

EIN=Level in coupler with no input minus RTG

SO…..

a) Present 60 dB SPL, determine average coupler SPL’s at 1000, 1600, and 2500 Hz, subtract 60 (ie RTG)

b) Remove the 60 dB SPL input and measure noise in the coupler

c) EIN= b-a

 

9) Harmonic Distortion

Occurs when new frequencies are generated that are harmonics of the original signal

VC in RTG position

Input = 70 dB SPL and measure total harmonic distortion at 500 and 800 Hz

Input = 65 dB SPL and measure total harmonic distortion at 1600 Hz

10) AGC Measurement

Static AGC Characteristics

Input-Output Curve

Input = measure SPL in coupler from 55 to 90 dB SPL in 5 dB steps

Dynamic AGC Characteristics

Input change jumping between 55 and 80 dB

Attack Time = time where level has stabilized to within 3 dB of steady state value for 90 dB signal

Release Time = time where level has stabilized to within 4 dB of steady state value for 55 dB signal

ANSI ‘96 Results

 

Time Constants (AGC Hearing Aids)

11) T-Coil Measures

Measured with a "wand" to simulate orientation to the phone

Measured across frequencies rather than at just 1 kHz

New terms:

SPLITS

HFA SPLITS

TMFS

SPLITS

SPLITS is an abbreviation for SPL in an inductive telephone simulator.

A coupler curve is run with the hearing aid in telecoil mode with the gain control at the reference-test position.

The input is a magnetic field generated by a Telephone Magnetic Field Simulator (TMFS).

HIGH-FREQUENCY AVERAGE SPLITS (HFA-SPLITS)—

The HFA-SPLITS value is the average of the dB levels at 1000, 1600, and 2500 Hz

taken
from a SPLITS curve with the hearing-aid gain control at the reference-test position.

TELEPHONE MAGNETIC-FIELD SIMULATOR (TMFS)

It is a device used to produce a magnetic field that is consistent in both level and geometric shape.

According to the standard, the current shall be equal to 6 milliamperes divided by the number of coil turns.

 

SUMMARY OF TOLERANCES

TEST TOLERANCE

OSPL90 Max Specified value + 3 dB

OSPL90 (HFA or SPA) Specified value ± 4 dB

Full-on gain (HFA or SPA) Specified value ± 5 dB

Reference-test gain None (information purposes only)

Reference-test position Gain control must be within ± 1 dB of the

target setting, such that the HFA

(or SPA) output for a 60 dB-SPL: input is 17 dB below the OSPL90 HFA (or SPA),

unless the full-on gain is already

lower

Frequency response curve Low band: specified curve ± 4 dB;

high band:specified curve ± 6 dB; curve may be shifted 10% left or right, and

unlimited up or down

 

SUMMARY OF TOLERANCES
TEST TOLERANCE

Frequency range None (info purposes only)

Percent total harmonic distortion Max is specified value + 3%

Equivalent input noise (EIN) level Max is highest specified value

+ 3 dB

Battery current drain Max is highest specified value + 20%

Input-output (I/O) characteristic After normalizing at 70dBSPL

input point, outputs for 50- and 90-dB-SPL inputs must be

within ± 5 dB or specified values.

Attack and release times Specified value ± 5 ms or ± 50%, whichever is larger

HFA- or SPA-SPLITS Within ± 6 dB of specified value

Simulated Telephone Sensitivity None (for info purposes only)

So ANSI S3.42…
are you comfortable with it?

Never be afraid to try something new.
Remember,

….Amateurs built the ark.

..... Professionals built the Titanic!

ANSI S3.42-1992
Testing Hearing Aids with a Broad-Band Noise Signal

Designed to test non-linear Hearing Aids with Broadband signals

Gives Family of 4 Response Curves

Inputs range from 40 to 90 dB SPL

ANSI S3.42-1992
Testing Hearing Aids with a Broad-Band Noise Signal

Reference:

Complex and Pure-tone Signals in the Evaluation of Hearing Aid Characteristics

Stemachowicz, P., Lewis, D., Seewald, R., and Hawkins, D.

JSHR 33, 380-385

ANSI S3.42-1992
Definitions

NSPL90

The root mean square (RMS) output sound pressure level (SPL) produced by a hearing aid with its gain set at maximum and with a 90 dB RMS speech-weighted noise input.

Standard measurement bandwidth is limited to 200 to 5000 Hz…

Full On Noise Gain

With the gain control on the hearing aid in the full-on position, the noise gain is a single figure expressed in dB and is derived by subtracting the overall RMS input SPL to the hearing from the RMS output SPL produced by the hearing aid

E.g., with a 50 dB RMS noise input, noise gain = output SPL – 50 dB

Root Mean Square

 

 

 

RMS = x12 + x22 + x32 +… xn2

n

RMS Calculation

ANSI ‘92 Results

ANSI ‘92 Results

Hearing Aid Settings

Adjust controls to the settings stated by the manufacturer

ANSI S3.42-1992
Testing Hearing Aids with a Broad-Band Noise Signal

Last Minute Advice to get this all done….


If you have a lot of tension and you get a
headache, do what it says on the

aspirin bottle:
"Take two aspirin"

and
 

EVALUATION PROCEDURES FOR NEW TECHNOLOGY

1) DIGITAL AIDS

2) MULTIPLE CHANNELS

3) ADAPTIVE RELEASE TIME

4) DIRECTIONAL MICROPHONES

5) GROUP DELAY AND PHASE

6) REDUCED RESIDUAL CANAL VOLUME (CICs)

1) EVALUATING DIGITAL AIDS

REVIEW OF : Digital vs. Analog

"Digital" sound…

Sound is "quantized" or described in very minute detail using a numeric code of 1’s and 0’s

This number code uses digits, hence the term DIGITal

"Analog"

The word "analogy" and "analogous" have same root

"Similar or alike enough to be compared"

Persons voice on the phone ANALAGOUS to their real voice…

Electroacoustic Tests with
Digital Aids

Testing noise reduction:

If tested with traditional complex steady-state signal, the aid interprets this as noise and reduces the gain

But when tested with interrupted signal, gain is not reduced

Important to test this function

EVALUATION OF NOISE REDUCTION CIRCUIT

BY USING THE DIGITAL SPEECH IN NOISE TEST (Star Option) CAN SEE THE EFFECTS OF SPECIAL NOISE REDUCTION CIRCUITS

also compare results to tests with

NOISE REDUCTION OFF……

Some Aids have
Test/Verification Mode

Phonak – Claro

Siemens – Signia, Prisma

Resound – 5000 Series

Sonic Innovation - Natura

Examples of DSP Hearing Aids Where a Modulated Signal May or May Not Be Required to Obtain Accurate REAG/REIG Measures.

Sonic Innovations: No

ReSound 5000/Canta: No

Widex Senso/Diva: Yes

Phonak Claro: Yes

Entering DSIN

The DSIN Screen

Two Speech Spectra

ANSI and ICRA Spectrum
(Frye, 2000)

Same Hearing Aid Measured With Three Signals
(Frye, 2000)

Adding a Bias Tone


ANSI Testing with a Bias Tone

No Bias and 500 Hz Bias Added in Between the Bursts of the Digital Speech (Frye, 2000)

No Bias and 4000 Hz Bias Added in
Between the Bursts of the Digital Speech
(Frye, 2000)

Some Aids have
Test/Verification Mode

Phonak – Claro

Siemens – Signia, Prisma

Resound – 5000 Series

Sonic Innovation - Natura

Phonak - Claro

GN Resound

Siemens Prisma

Siemens - Signia

Sonic Innovation

EVALUATION PROCEDURES FOR NEW TECHNOLOGY

1) DIGITAL AIDS

2) MULTIPLE CHANNELS

3) ADAPTIVE RELEASE TIME

4) DIRECTIONAL MICROPHONES

5) GROUP DELAY and PHASE

6) REDUCED RESIDUAL CANAL VOLUME (CICs)

2) EVALUATION OF MULTIPLE CHANNELS

Many programmable aids have more than a single channel

Automatic signal processing may result in different responses in different channels

With ANSI S3.42 ’92 can see the gain effects across the channels

With "Enhanced Attack and Release" can see the temporal effects across channels

CHECK FREQUENCY RESPONSE

ANSI S3.42 ’92 – to see differential changes in frequency response across channels

CHECK ATTACK/RELEASE TIMES

CHECK
INPUT/OUTPUT CURVES

I/O CURVES

A choice of test frequencies to measure I/O curves in frequency regions where different compression thresholds and compression ratios have been set

 

 

 

 

Inquiring

Beavers

want to know!!!

 

 

 


Beavers want to know…..

What happens in different channels?

What does the noise reduction do?

How much difference does the directional microphone make?

EVALUATION PROCEDURES FOR NEW TECHNOLOGY

1) DIGITAL AIDS

2) MULTIPLE CHANNELS

3) ADAPTIVE RELEASE TIME

4) DIRECTIONAL MICROPHONES

5) GROUP DELAY and PHASE

6) REDUCED RESIDUAL CANAL VOLUME (CICs)

3) ADAPTIVE RELEASE TIME

Some circuits designed with release times that vary with the duration of the input signal.

This is intended to activate short RTs for sounds such as door slams but longer RTs for continuous noise to prevent it from being amplified during pauses in speech

ADAPTIVE RELEASE TIME

The Adaptive Attack &Release Test performs a 2000-Hz attack test,followed by a special,two-signal release test consisting of a

short signal (1/10 second)and

long signal (2 seconds)

You can select whether the signal level varies between 55 and:75,80,85,90,or 95dB SPL

(80 is standard).

ADAPTIVE RELEASE TIME

EVALUATION PROCEDURES FOR NEW TECHNOLOGY

1) DIGITAL AIDS

2) MULTIPLE CHANNELS

3) ADAPTIVE RELEASE TIME

4) DIRECTIONAL MICROPHONES

5) GROUP DELAY and PHASE

6) REDUCED RESIDUAL CANAL VOLUME (CICs)

4)DIRECTIONAL MICROPHONES

It is possible that by just using conventional verification techniques (real ear measures) that malfunctioning directional mics could be missed

Helpful to demonstrate to the patient the benefit of directional mics

MICROPHONE TYPES

Omni-directional

Directional

Dual Microphone

Dynamic microphone designs

Multiple mic arrays

endfire (along side of head)

broadside (across the front)

MEASURE FRONT TO BACK RATIO

Measure real ear performance

with speaker in the front and in the back of the patient

to verify the rear mic is actually working to reduce sounds from the rear

FRONT TO BACK RATIO MEASURES

1) dB SPL plots

loudspeaker in front

loudspeaker in rear

2) Gain plots

loudspeaker in front as "REUR"

loudspeaker in rear as "REAR"

"REIG" is NOW the…..

"Front-to-Back Ratio"

3) Can also measure in the test box by rotating the aid in the box with the lid open

How are we doing???
CLUELESS???

EVALUATION PROCEDURES FOR NEW TECHNOLOGY

1) DIGITAL AIDS

2) MULTIPLE CHANNELS

3) ADAPTIVE RELEASE TIME

4) DIRECTIONAL MICROPHONES

5) GROUP DELAY AND PHASE

6) REDUCED RESIDUAL CANAL VOLUME (CICs)

Enhanced DSP Option

Designed to measure the "group delay" of a DSP aid. In other words, the actual digital processing time of a hearing aid.

0 ms for analog aids

1-6 ms (or greater) for DSP

Easy way to "see" of aid is analog or DSP

Also tests the "phase" of a hearing aid. This is useful for binaural fittings. Are the aids working together?

Technical Details
Group Delay

Technical Details
Phase measurement

Entering Enhanced DSP

Enhanced DSP Test #1

Phase comparison

Enhanced DSP Test #2

Enhanced DSP Test #3

Enhanced DSP Test #4

Phase comparison

Enhanced DSP #5

Phase Comparison

Enhanced DSP #6

Comparison of Phase

New Phase Differences Discovered

EVALUATION PROCEDURES FOR NEW TECHNOLOGY

1) NOISE REDUCTION

2) MULTIPLE CHANNELS

3) ADAPTIVE RELEASE TIME

4) DIRECTIONAL MICROPHONES

5) GROUP DELAY AND PHASE

6) REDUCED RESIDUAL CANAL VOLUME (CICs)

6) REDUCED RESIDUAL CANAL VOLUME (CICs)

The size of the residual ear canal volume is much smaller with CIC than with other aids….

BTE, ITE, and CIC’s tested with 2cc coupler

CIC tested with new coupler- .4cc

REDUCED RESIDUAL CANAL VOLUME (CICs)

Not intended to replicate an ANSI run

Two options:

Manual Mode

Automatic Mode

Attach aid to CIC coupler in the same manner as ITE is attached to HA1

REDUCED RESIDUAL CANAL VOLUME (CICs)

Compare differences re: HA2 and CIC coupler

First test with HA2

Next test with CIC

The greater output with CIC coupler is more realistic of what patients receive

REDUCED RESIDUAL CANAL VOLUME (CICs)

In Manual Mode………

Test as desired…

Multi-Curve,GAIN,

Digital Speech-in-Noise, Profiler, and AVG options

In Automatic Mode…..

Same tests as in ANSI ’87

Can choose options in CIC Setup menu

REDUCED RESIDUAL CANAL VOLUME (CICs)

AUTO CIC SETUP

AID TYPE LINEAR; AGC WITH EIN; AGC WITHOUT

EIN; ADAPTIVE WITH EIN; ADAPTIVE AGC

WITHOUT EIN

FULL ON GAIN 60 dB; 50 dB

TELECOIL DISABLED; ENABLED

AVERAGE FREQUENCIES 1000 1600 & 2500 Hz HFA;

1250 2000 & 3150 Hz (SPA);

1600 2500 & 4000 Hz (SPA);

2000 3150 & 5000 Hz (SPA);

800 1250 & 2000 Hz (SPA)

DELAY MENU

REDUCED RESIDUAL CANAL VOLUME (CICs)

SUMMARY

VERIFICATION OF NEW TECHNOLOGY EXTREMELY IMPORTANT….

SUPPORT THE NEED FOR ADVANCED FEATURES

VERIFY FOR THIRD PARTY PAYMENT

DEMONSTRATION TO PATIENTS

OPTIMIZE FITTING PROCESS

SUMMARY

1) NOISE REDUCTION

2) MULTIPLE CHANNELS

3) ADAPTIVE RELEASE TIME

4) DIRECTIONAL MICROPHONES

5) GROUP DELAY and PHASE

6) REDUCED RESIDUAL CANAL VOLUME (CICs)

SUMMARY

DON’T BELIEVE THE MYTH THAT DIGITAL HEARING AIDS CANNOT BE TESTED…..

ONLY TRUE IF STUCK IN MODE OF PUSHING BUTTON FOR ANSI AUTOMATED RUNS

BREAK AWAY FROM TRADITIONAL AUTOMATED RUNS

EXAMINE YOUR INSTRUMENTS!!!

Our Core Questions:

How well do the simulated 2-cc coupler response curves agree with measured curves?

How well do the attack / release-time data provided agree with these time constants as measured re: ANSI S3.22-1996?

3 levels on screen "Simulated 2cc Output"

Procedure

BTEs were left connected to Hi Pro box throughout programming & measurement

BTEs were programmed for 4 different hearing losses

Simulated frequency responses

Data Analysis

Simple difference score:

Simulated minus 2cc-coupler

RMS difference score:

RMS = x12 + x22 + x32 +… xn2

n

Simulated – 2cc-coupler Response

Another Example: (Simulated – 2cc)

Results

Average differences, all BTEs

…So, what you SEE is what you GET?

The average difference data are just that…averages

The range of difference scores is important

Following are a few examples…

Average for the Loss

The aid with the smallest difference scores

The aid with the largest difference scores

Loss A

Loss B

Loss C

Loss D

RMS Calculation

…from previous slide: 6,7,3,3,3,3

Square values & sum: 36+49+9+9+9+9=121

Get mean value: 121¸ 6 = 20.167

Take the square root: Ö 20.167 = 4.5

Another Example: (Simulated – 2cc)

RMS calculation (Sim. – 2cc Coupler)

RMS = Ö [(1+25+4+1+16+4)¸ 6]=2.9

 

Measures of Attack and Release Time

Some preliminary data

BTE #5: the manufacturer’s software provided attack and release time measures in the low and high channels




 

 

 

Summary and Conclusions

On average, across a wide range of input levels, hearing aids, and hearing losses, simulated estimates of 2cc-coupler responses are within ± 2 dB of measured values

Significant departures (> ± 5 dB) from the predicted 2cc-coupler values occurred 27.5% of the time

(192 / 696 measures)

Summary and Conclusions, continued

Differences between digital vs. analog? …not significant

These findings reaffirm the utility of obtaining 2cc-coupler measures prior to fitting the hearing aids

Summary and Conclusions, continued

Release times as measured using ANSI S3.22-1996 varied from those shown by the software

Here, our n=1…

Comparative research is continuing

Recommendations…

Thank you for your Attention!


SIARC

Summer Intensive Aural Rehabilitation Conference

Weeklong Intensive Program for communication partners to learn communication strategies, coping skills, and use of ALDs in the real world!
July, 2004

In Summary…….

 

 

 

 

 

Input vs. Output Compression

5) INTERNAL or BOOT
FM RECEIVERS

 

BEHIND THE EAR FM RECEIVERS

Telex Select 1-40

Phonic Ear Free Ear

BOOT FM RECEIVERS

Phonak MicroLINK

 

 

 

Testing with Direct Audio Input

Testing with Neckloop

Testing Environmental Microphone

DEMO- MEASUREMENTS WITH FM SYSTEMS

 

REAL EAR MEASUREMENTS WITH FM SYSTEMS

 

REAL EAR MEASUREMENTS WITH FM SYSTEMS

 

REAL EAR MEASUREMENTS WITH FM SYSTEMS

 

Electroacoustic Tests with
Digital Aids

Also need to test channel functions

Frye 6500 test box allows for a "bias" tone

The "bias" tone is interpreted as noise in just that frequency region and a filter in that region is initiated