audiogram

Thread view:

Enable EMail Alerts

Start New Thread

Thread rating:

Tom van Grootel - 25 Dec 2003 03:03 GMT

When testing a person's/patient's sense of hearing in general a method is
used where this subject has to verbally indicate when a tone stimulus can
be heard. These tone stimuli are usually presented sequentially from loud
to soft (in steps of a few dBs) in one frequency. When the threshold is
determined, the next frequency is tested.

I am a bit bothered by this method although it might be fast, because
subjects can falsely indicate the stimulus as "heared", because they might
know the rate of succesion of the stimuli. Also they can miss a stimulus
easily when choughing or by another disturbance.

my question: is there another method known, which gives a good indication
of the sense of hearing?

my intuition: some method with a 'forced choice' instruction, where in 50%
of the cases a stimulus is presented, without succesion and not at fixed
frequencies. The results can be interpreted as follows: 50% right, at
threshold (equal to 'quessing'), >50% right, above threshold, <50% right,
below threshold, and extrapolated between the (not discrete!) frequencies.

Thanks for your critical responses
Tom

Reply to this Message

Eric Desart - 25 Dec 2003 20:12 GMT

Hello Tom,

There are better experts here in this field.

What however I notice in practice is that the responsible one will check in
both directions. From loud to silent, and then plays a bit around this
bottom level, until he/she is pleased that the subject indeed responded
correctly.

But I agree, it still leaves room for error.

Eric

> When testing a person's/patient's sense of hearing in general a method is
> used where this subject has to verbally indicate when a tone stimulus can
[quoted text clipped - 18 lines]
> Thanks for your critical responses
> Tom

Reply to this Message

Angelo Campanella - 25 Dec 2003 22:39 GMT

> I am a bit bothered by this method although it might be fast, because
> subjects can falsely indicate the stimulus as "heared", because they might
> know the rate of succesion of the stimuli. Also they can miss a stimulus
> easily when choughing or by another disturbance.

    The short answer is "Bekesey had it right".

    One of the original audiometrs, certainly the most popular compeate
audiometer of its time was the Bekesey Audiometer, popular in the 1940's
and 1950's. This unit continuosly scanned a tone while the subject held
a pushbutton switch. A chart recorder continuosly displayeed and
recorded the tone level. On startup, the tone level was automaticlly
increased at a steady rate. When the button was pressed, the sound level
was diminished. One held down that button until the tone was not
audible, then released it, whence the tone level increase again, etc.
    The tone frequency scan rate was slow enough so that the recorded trace
became a narrow sawtooth whose span bracketed the threshold level. The
lowest test frequency was around 100 Hz, and the highest at least 10,000
Hz. A beautiful and comletely informative audiogram trace, no frequency
gaps, was the result. (I wish I has preserved the one done on me then,
but that was not to be.)

    In successive years the industry "value engineers" re-engineered
audiometers to be "cost effective", and this public art was lost. Today,
with so much systems enineering going on, undoubtedly, the Bekesey
system will soon be reinvented. And it will be none too soon.

    Angelo Campanella

Reply to this Message

Peter Larsen - 26 Dec 2003 12:10 GMT

> When testing a person's/patient's sense of hearing in general a method is
> used where this subject has to verbally indicate when a tone stimulus can
> be heard. These tone stimuli are usually presented sequentially from loud
> to soft (in steps of a few dBs) in one frequency. When the threshold is
> determined, the next frequency is tested.

> I am a bit bothered by this method although it might be fast,

There is a psychologic aspect you have overlooked: the desire to perform
well when challenged and a psychoacoustic one: it is very difficult to
fail the test on purpose in a non-obvious way. That said, it indeed
appears that Angelo yet again got it right: Bekesy got it right.

> Tom

Kind regards

Peter Larsen

Signature

***************************************************************
* \\\\\\\ Quality Ascii handcrafted by Peter Larsen /////// *
* \\\\\\\ My site is at: http://www.muyiovatki.dk /////// *
*********************************************************

Reply to this Message

Tom van Grootel - 28 Dec 2003 17:08 GMT

hi Eric, Angelo, Peter and Elly

Thanks for your reactions.

> That said, it indeed appears that Angelo yet again got it right: Bekesy
> got it right.

Actually i was expecting a bit more opposition, because this method of
measuring sense of hearing is a largely used and perhaps considered as
sacred. Nevertheless you agreed that things can be improved, or at least
there is a better method.

In the output of the Bekesy audiometer:
http://facstaff.uww.edu/bradleys/ohc/audiometry/audio_bekesy.gif
you can clearly see the hearing threshold is not determined as a sharp
line. The 'sawtooth' line gives this away, with the release and
press of the button being the extremes.

What i wanted to raise is a discussion that in assesing sense of hearing
one should carefully consider the error of measurement.

kind regards, Tom

Reply to this Message

Peter Larsen - 28 Dec 2003 23:16 GMT

> What i wanted to raise is a discussion that in assesing sense of hearing
> one should carefully consider the error of measurement.

A good audiogram is well within the confines of a 20 dB wide zone of
personal variance, even just taking a city dweller to the quiet country
side for 3 weeks could alter the overall sensitivity by 10 dB.

> kind regards, Tom

Signature

Reply to this Message

Malcolm Hayes - 29 Dec 2003 10:03 GMT

Have you a reference for this statement? (The 3 weeks and 10 dB change in
sensitivity)

MalcolmX

> A good audiogram is well within the confines of a 20 dB wide zone of
> personal variance, even just taking a city dweller to the quiet country
[quoted text clipped - 7 lines]
> * \\\\\\\ My site is at: http://www.muyiovatki.dk /////// *
> *********************************************************

Reply to this Message

Kari Pesonen - 29 Dec 2003 12:42 GMT

> Have you a reference for this statement? (The 3 weeks and 10 dB change in
> sensitivity)
[quoted text clipped - 6 lines]
> >
> > > kind regards, Tom

The question concerns recovery of Noise Induced Temporary Tresheold
Shift. Recovery time (to reach a constant hearing threshold) depends on
the original noise dose (that produced the TTS) and also on the noise level
during revovery. To have any recovery the sound level should be bellow
the level (daily dose level) of effective silence (ca. 77...78 dB(A))
Have a look, for example,

Miller J, Effects of noise on people, J. Acoust Soc. Am. 56(1974)3, 729 – 764.

Melnick W, Human asymptotic threshold shift, in Henderson D et al., (eds.),
Effect of noise on hearing, Raven, New York 1976, ss. 277 – 289.
Melnick W, Human temporary threshold shift (TTS) and damage risk,
J. Acoust. Soc. Am. 90(1991)1, 147 – 154.

Kari Pesonen

Reply to this Message

Angelo Campanella - 29 Dec 2003 00:12 GMT

> What i wanted to raise is a discussion that in assesing sense of hearing
> one should carefully consider the error of measurement.

    There is a more fundamental problem regarding hearing sensitivity
assessment:

    Should we be pursuing the treshold instance, or should we rather pursue
the level where 50% probability of detection occur, or 90% or 99%, etc.?

    Should we revisit the levels of loudness perceived by humans with
average nearing up to, say, the "extremely loud" level (viz 100 dB)?

    Regarding your error of measurement proposal, the classical method of
determining that is to perform controlled experimental measurements of a
specific status of perception. You are welcome to propose (and perform
if you wish) crtiteria of human detection (perception threshold) for
sound of various frequencies.

    Angelo Campanella

Reply to this Message

Eckard Blumschein - 29 Dec 2003 09:16 GMT

> Should we revisit the levels of loudness perceived by humans with
> average nearing up to, say, the "extremely loud" level (viz 100 dB)?

With very low frequency some people might actually perceive sound at the
threshold of hearing like extremly loud and annoying. This was at least
one attempt to explain notorious complaints. Present standards do not
sufficiently consider such aspects.

www.lfu.baden-wuerttemberg.de/lfu/abt3/brummton/ergebnis.html shows
audiograms which were carefully recorded down to 25 Hz. They exhibit
large individual differences. The subjects were affected from the so
called hum. However, the audiograms could not explain this phenomenon.
Nonetheless, they show a zigzag only down to about 100 Hz.

Sensation of low frequency seems to depend on neural peculiarities and
periodicity rather than spectrally resolving hair cells with
corresponding characteristic frequency (CF).
Perhaps the lowest CF is in the ballpark of 140 to 70 Hz. Perception of
frequencies as low as 20 Hz or less does obviously not benefit from the
cochlear amplifier. So it requires a SPL in excess of 70 dB.

Taking such concerns into account, I would not exclude the possibility
to improve audiography. However, more intelligent systems could be
differently designed for different tasks. BTW, does the audiology news
group still exist?

Eckard Blumschein

Reply to this Message

Tom van Grootel - 29 Dec 2003 13:47 GMT

Hi Angelo,

> You are welcome to propose (and perform if you wish) crtiteria of human
> detection (perception threshold) for sound of various frequencies.

I don't have the resources to perform these kind of experiments (working
on the visual system at the moment). But I was thinking of a simple
experiment, which i described before, and didn't have the time to
carefully think it over:

Method with a 'forced choice' instruction, where in less than 100% of the
cases a stimulus is presented, without succesion in frequency or intensity
The results can be as follows:
50% right, (equal to 'guessing') threshold (or is 99% threshold ???)

>50% right, above threshold ???,

<50% right, below threshold ???,
and extrapolated between the (not discrete!) frequencies.
This can be tested with different subjects and in different sessions.

good idea? (BTW i am not a great expert)

Tom

Reply to this Message

Angelo Campanella - 29 Dec 2003 22:03 GMT

> Method with a 'forced choice' instruction, where in less than 100% of the
> cases a stimulus is presented, without succesion in frequency or intensity
> The results can be as follows:
> 50% right, (equal to 'guessing') threshold (or is 99% threshold ???)

Thank you for the suggestions. But there is a more fundamental problem,
where threshold determination methods are not as important as the
variation of personal thresholds, and the condition of the persons at
the time of testing.

The answer by Eckard Blumschein is very good.

We should spend some time sudying the Baaden-Wuerttemburg reslts.
The low frequency determinations show a wide spread, depending on such
conditions

Angelo Campanella.

Reply to this Message

Natural Science Forum / Physics / Acoustics / December 2003

audiogram