"Theshold levels of human perception"
James Boyk
> Nonsense! Today's well-equipped audio labs can measure all of the
> audible (and other) properties of cables at levels very, very much
> lower than the threshold levels of human perception.
To be able to say this, you must feel confident that you know what
the relevant human perceptions are and what is the threshold level of
each. Would you kindly specify both, and cite your authorities?
Stephen E. Halpin
>DALJHD wrote [in another thread]:
>
>> Nonsense! Today's well-equipped audio labs can measure all of the
>> audible (and other) properties of cables at levels very, very much
>> lower than the threshold levels of human perception.
>
>To be able to say this, you must feel confident that you know what
>the relevant human perceptions are and what is the threshold level of
>each. Would you kindly specify both, and cite your authorities?
Ive been following this newsgroup and the assorted trade rags for
some time now, and I think its about time that consumers push some
of the burden of proof on the manufacturers and the trade rags.
One the one hand, the rags seem to go on and on about how the
current piece under test is one of the best examples of X that
they have ever heard, and comperable to a unit costing N times
as much. By induction, almost any X at any price is as good
as the baseline equipment that the rag will review. At the
same time, the measurements that are taken (noise floor,
frequency response, impedance, etc...) are often what one might
consider poor, and yet if the reviewer liked the equipment they
write it off as "inaudiable." Given how many things fall into
this "inaudiable" catagory, its amazing that anyone associated
with this industry says that small measurable differences are
significant.
On the other end of the spectrum are the manufacturers who
publish "measurements", some of which are obviously drawn
with a crayon rather than a measuring device. There is one
web site which shows how their product improves sound, and
yet when you look at the >>wildly<< different waveforms (keep
in mind, these are no where near as fine grained as the sample
level) one would have to believe from the radically different
amplitude and timing differences that one could put a Spice Girls
CD in and get the Boston Symphony Orchestra playing an entirely
different piece out the other side. Anything Ive seen that
affects a digital bit stream is supposed to affect a specific
frequency range, and yet to do that it would have to have digital
processing capabilities far beyond what a wire or magnet can
provide. I have never seen bit error rate numbers. I have
never seen cable ads which show how their minor impedance tuning
really makes a measurable >sonic< difference against a highly
complex, significantly larger impedance speaker.
Its time that meaningful, objective measurements are developed,
and if the equipment today isnt measuring the right stuff, some
research should be going towards finding and implementing those
measurements. It would also help if manufacturers published
any meaningful specifications. You can find literature for
$15,000 amps which wont even provide an impedance with the one
listed power rating...
If it sounds like I am frustrated by what I read, I am. I
personally think too many live in the land of make believe,
wanting to be part of the club rather than experiencing better
sound. Hi end audio has been around for decades, and its about
time that the consumer be given some reasonable facts to work
with. Most stores only carry one or two brands at a price
point, which makes it impossible to hear two similar units
back to back for a fair comparison. To make matters worse,
room setup plays a big difference and just going from one
room to another makes a meaningful comparison all but impossible.
For the systems I have looked at, more money buys a different
set of tradeoffs, but the overall sound is often not "better."
How are we supposed to make good decisions in this environment?
>James Boyk
>http://www.cco.caltech.edu/~boyk/records.html
-Steve
Richard D Pierce
James Boyk <bo...@caltech.edu> wrote:
>DALJHD wrote [in another thread]:
>
>> Nonsense! Today's well-equipped audio labs can measure all of the
>> audible (and other) properties of cables at levels very, very much
>> lower than the threshold levels of human perception.
>
>To be able to say this, you must feel confident that you know what
>the relevant human perceptions are and what is the threshold level of
>each. Would you kindly specify both, and cite your authorities?
Starting with Hermann von Helmholtz and moving through Fletcher,
Munson, and others in the 150 or so years of such research, whch
of the hundreds upon hundreds of studies would you like?
You aren't SERIOUSLY suggesting that these thresholds are not
well studied and the researchers are not well known, are you?
--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| DPi...@world.std.com |
joh...@attglobal.net
standardized measurements going to be useful?
I sympathize with your frustration though :-)
JohnR
Richard D Pierce
>If room setup makes such a huge difference, then how is a set of
>standardized measurements going to be useful?
Your premise makes some assumptions that are false. If the
ssumption that room setup makes a huge difference is valid, that
would predict that differences between speakers are overwhelmed
by differences in rooms and setups, and this is known to be
false.
Rooms CAN make a significant difference in low frequencies. But
consider the fact that at higher frequencies (often as low as
200 Hz), the absorbtion coefficient of room furnishings is high
enough that the room has a far smaller effect than assumed.
Secondly, much of the information that is used to clue the ear
into the sonic characterstics of speakers is gained in the
direct arrival sound, which is utterly unaffected by the room
(byt definition: if it is direct arrival, the room cannot
possibly have affected it).
One finds, contrary to your assumption, that the frequency
response of MOST loudspeakers in real rooms above 500 Hz does
NOT differ markedly from the anechoic response.
>I sympathize with your frustration though :-)
But not with incorrect assumptions, it is hoped.
Art Ludwig
>
>One finds, contrary to your assumption, that the frequency
>response of MOST loudspeakers in real rooms above 500 Hz does
>NOT differ markedly from the anechoic response.
By coincidence I just measured this, and in fact there is a
substantial room effect at high frequencies. The results below are
obtained by taking a single impulse measurement using the CLIO MLS
system. I first set the time gate at the maximum value, about 1/3
second, to capture all of the significant room reflections. CLIO does
the FFT and I export the result as an ascii file. I then reset the
time gate to gate out all room reflections, do another FFT & export.
The dB SPL values are subtracted to get the plots. The result is the
change in frequency response produced by the room. This procedure has
the virtue that it is independent of the speaker frequency response,
and yields the changes that would be produced for any speaker with
the same directivity pattern. The smaller time gate makes the data at
low frequencies invalid, but above about 500 Hz it is accurate. My
room is 14 x 21.5 x 8 feet in size, heavy carpet on the floor,
wallboard walls and ceilings. There are two large upholstered chairs
and misc other furniture. There are also foam sound absorbing panels
as noted below. In another forum someone suggested that a line source
speaker radiating a dipole front/rear pattern would have a smaller
effect than cone speakers. I doubted that, and to settle the question
measured both. The line source is a Acoustat 1+1 floor to ceiling,
set 3-feet from the rear wall, as recommended by a manufacturer of
such speakers. The results can be seen at
www.silcom.com/~aludwig/images/Reflection_effect_linesource_dB.gif
There are changes in excess of 10 dB. The upward bias in the change
is due to "room gain" (see my site at www.thesoundpage.com for
details) For my main speaker system, 3-way with cones/dome, I put a
lot of effort into reducing early room reflections, including 4 large
foam absorbing panels. Again details can be found on my site. The
room effect for this system can be seen at
www.silcom.com/~aludwig/images/Reflection_effect_3way_dB.gif
The variations are substantially reduced, but still significant. My
guess is that these two cases probably are close to the worst and the
best that can be expected.
Regards, Art Ludwig
KlausRa...@compuserve.com
bo...@caltech.edu wrote:
> DALJHD wrote [in another thread]:
>
> > Nonsense! Today's well-equipped audio labs can measure all of the
> > audible (and other) properties of cables at levels very, very much
> > lower than the threshold levels of human perception.
>
> To be able to say this, you must feel confident that you know what
> the relevant human perceptions are and what is the threshold level of
> each. Would you kindly specify both, and cite your authorities?
>
> http://www.cco.caltech.edu/~boyk/records.html
>
I recently posted on the cable asylum what answers your question.
http://www.audioasylum.com/audio/cables/messages/6127.html
There you find some bibliographic data of documents clearly indicating
that cable sound is pure aural illusion.
klaus
Sent via Deja.com http://www.deja.com/
Before you buy.
KlausRa...@compuserve.com
psychoacoustics was made when Charles Lindberg was
coming back from his Atlantic flight and was
celebrated by the city of New York. A scientist then
tried to figure out the masking effect of the crowd's
noise with respect to tones produced by headphones.
I'm not quite sure about that, but I think he
published his results in 1929 in JASA (Journal of the
Acoustical Society of America).
I think it was also him who brought up the concept of
critical bands.
Klaus
James Boyk
Nonsense! Today's well-equipped audio labs can measure all of the
audible (and other) properties of cables at levels very, very much
lower than the threshold levels of human perception.
James Boyk wrote: To be able to say this, you must feel confident
that you know what the relevant human perceptions are and what is the
threshold level of each. Would you kindly specify both, and cite your
authorities?
KlausRa...@compuserve.com wrote:
I recently posted on the cable asylum what answers your question.
James Boyk writes:
Sorry not to make myself clear? An answer to my question (above) will
consist of...
1. An explicit list of human perceptions relevant to perceived cable
sound quality, with citations to the experiments that established
their relevance. (If desired, the perceptions could be those relevant
instead to the sound quality of any other type of component.) This
list will not consist merely of citations to the scientific
literature. The original claim was implicitly of knowledge of "all of
the audible...properties of cables." To support this claim, those
properties must be explicitly listed and proved to be relevant by
existing scientific papers. (At least some of these papers must of
course report on detailed listening tests on cables, or whatever type
of component is in question.)
2. The scientific literature referred to in the list should also
establish thresholds for perception of each of the "audible
properties."
This request is a reasonable & scientific one, given the original
claim.
James Boyk
Richard D Pierce
James Boyk <bo...@caltech.edu> wrote:
>James Boyk wrote: To be able to say this, you must feel confident
>that you know what the relevant human perceptions are and what is the
>threshold level of each. Would you kindly specify both, and cite your
>authorities?
>
>KlausRa...@compuserve.com wrote:
>I recently posted on the cable asylum what answers your question.
>
>James Boyk writes:
>
>Sorry not to make myself clear? An answer to my question (above) will
>consist of...
>
>1. An explicit list of human perceptions relevant to perceived cable
>sound quality, with citations to the experiments that established
>their relevance. (If desired, the perceptions could be those relevant
>instead to the sound quality of any other type of component.) This
>list will not consist merely of citations to the scientific
>literature. The original claim was implicitly of knowledge of "all of
>the audible...properties of cables." To support this claim, those
>properties must be explicitly listed and proved to be relevant by
>existing scientific papers. (At least some of these papers must of
>course report on detailed listening tests on cables, or whatever type
>of component is in question.)
It's really VERY simple.
We know, with great confidence, what the threshold of human
hearing is to signals. We know, for example, that people CANNOT
hear ANYTHING below a sound pressure 0.0002 dynes/cm^2 around 1
kHz. Not a single shred of evidence from ANYONE suggests
otherwise. For your references, I would start with Fletcher and
Munson. And search all you want, you will find no contradictory
references.
So we can say, with great confidence, that if a component
generates residuals the effect of which is a 1 kHz component
that results in a sound pressure level AT THE EAR that's less
than 0.0002 dynes/cm^2, it WILL NOT BE DETECTED. And this is
further supported by the plain and simple fact that under even
the most optimum music listening conditions, the actual noise
level is FAR in excess of that 0.0002 dynes/cm^2, so now we have
masking effects to contend with as well.
The residuals generated by cables are at these levels and well
below. Unless you're willing to invoke psychic powers, there
does not exist a single shred of credible evidence that suggests
that these residuals cause any auditory stimulus whatsoever.
AudioMaven
the effect of which is a 1 kHz component
that results in a sound pressure level AT THE EAR that's less than 0.0002
dynes/cm^2, it WILL NOT BE DETECTED. And this is further supported by the plain
and simple fact that under even the most optimum music listening conditions,
the actual noise level is FAR in excess of that 0.0002 dynes/cm^2, so now we
have
masking effects to contend with as well.>>
Dear Dick,
I'm interested to know more about how these experiments were carried out and
the statistical make up of the test group.
How many people were involved? How many people would have to be involved for
the tests to statistically show that say 30% of the population could discern a
difference, not 100% (this is a big consideration that we run into when testing
new chemotherapy drugs for instance). I think we can say that everyone is
different genetically and comes with different hearing abilities (based upon
physiology say) -the best example being say those blessed with perfect pitch on
one extreme and those that are tone deaf on the other extreme.
How were the tests carried out? Headphones? What was the age, sex, ethnic
origin and occupations (again not insignificant consideration based on the
nerve ending studies of female vs. male populations) of the test population?
Myles B. Astor
Publisher
Ultimate Audio magazine
www.ultimateaudio.com
jj, curmudgeon and tiring philalethist
James Boyk <bo...@caltech.edu> wrote:
>1. An explicit list of human perceptions relevant to perceived cable
>sound quality, with citations to the experiments that established
>their relevance.
Fletcher's 0 Loudness curves.
That's it. End of sentence. Old work. Measurement equipment
is more sensitive than that. Period.
>2. The scientific literature referred to in the list should also
>establish thresholds for perception of each of the "audible
>properties."
There need be only one, that of basic, most-sensitive audible
threshold.
Equipment goes WELL below that.
Q.E.D.
--
Copyright j...@research.att.com 2000, all rights reserved, except transmission
by USENET and like facilities granted. This notice must be included. Any
use by a provider charging in any way for the IP represented in and by this
article and any inclusion in print or other media are specifically prohibited.
KlausRa...@compuserve.com
> Sorry not to make myself clear? An answer to my
question (above) will consist of...
> 1. An explicit list of human perceptions relevant
to perceived cable sound quality, with citations to
the experiments that established their relevance. (If
desired, the perceptions could be those relevant
instead to the sound quality of any other type of
component.)<
To establish such a list, one would have to know what
cable sound consists of. Exactly this is what all the
cables experts are silent about.
When checking the AES archive at www.aes.org (from
1953 to 1995, I checked the later volumes manually)
for cables you find papers on transmission lines in
studios, a paper about cables for digital signals,
and the speaker cable papers Davis and Greiner I
referred to. This also could be considered as
indication that short length cables for audio
frequencies are of no importance, otherwise they
would have been subject of more than just two papers
(the AES published about 450 papers on loudspeakers,
just to give an counterexample).
The papers by Greiner clearly show that the complex
load a cable represents is far smaller than the load
the crossover/voice coil system of the speaker
represents. An amplifier said to be sensible to such
cable load has problems with the speaker's complex
load anyway.
The 1991 AES paper by Davis measures amplitude vs
frequency (of commercialized cables connected to amp
and speakers) and presents results in order of
magnitude of (at worst) 0.3 dB down from flat at
about 20 kHz. Florentine 1987 found that for pure
tones at 16 kHz a deviation of about 2.0 dB is needed
to be audible (deviations needed becoming greater
with increasing frequency). Masking effects (upward
masking) when playing real music are not considered
in that paper.
Not known is whether cables induce phase shifts. But
even if, literature of threshold for both group delay
and delay between fundamentals and harmonics exist
(see my list) so that such cable phase shifts could
be checked for audibility.
Exotic phenomena (as listed in my posting) are found
and can be measured. But they are never quantified.
It remains thus open if they actually contribute to
cable sound.
>The original claim was implicitly of knowledge of
"all of the audible...properties of cables." To
support this claim, those properties must be
explicitly listed and proved to be relevant by
existing scientific papers.<
I'm no cable expert. Go to the cable asylum or to
rec.audio.tech and ask that question.
>(At least some of these papers must of course report
on detailed listening tests on cables, or whatever
type of component is in question.)<
Go to the DBT link in the cable asylum. It guides you
further to the ABX site where data on ABX-listening
tests with speaker cables are presented. I assume
that DBT on cables is avoided for the reason of being
afraid to find out that this all is an illusion.
>2. The scientific literature referred to in the list
should also establish thresholds for perception of
each of the "audible properties.">
Such thresholds are known (see my list). If "new"
audible porperties can be found, I've got more papers
to cite for such thresholds.
> This request is a reasonable & scientific one,
given the original claim.>
I was also interested in some background other than
"cable sound is reality, go and listen yourself".
That's why I dug a bit in related literature.
In my posting on the cable asylum I stated that
amplitude vs frequency and phase vs frequency were
the only parameters involved in cable sound. Since
nobody contested this and corrected that statement I
assume that there is no other relevant parameter.
Klaus
Ball hit back to the cable fraternity.
jj, curmudgeon and tiring philalethist
<KlausRa...@compuserve.com> wrote:
>To establish such a list, one would have to know what
>cable sound consists of. Exactly this is what all the
>cables experts are silent about.
Klaus, that is simply an erronious consideration.
The dispute over measurement is clear, the sensitivities
of the human auditory system to minimum stimulii are well
known and established, and confirmed to a fair-thee-well.
Instruments are more sensitive than the auditory system
in this respect, so it IS possible to measure cable
distortions, if any, more accurately than the auditory
system can detect them.
Now, if we CAN detect substantial distortions above that
level, it's a different question, but one that wasn't being
asked at the minute I looked in.
KlausRa...@compuserve.com
audio...@aol.com (AudioMaven) wrote:
> I'm interested to know more about how these experiments were
> carried out and the statistical make up of the test group.
> How many people were involved? How many people would have to be
> involved for the tests to statistically show that say 30% of the
> population could discern a difference, not 100% (this is a big
> consideration that we run into when testing new chemotherapy drugs
> for instance). I think we can say that everyone is different
> genetically and comes with different hearing abilities (based upon
> physiology say) -the best example being say those blessed with
> perfect pitch on one extreme and those that are tone deaf on the
> other extreme.
> How were the tests carried out? Headphones? What was the age, sex,
> ethnic origin and occupation (again not insignificant consideration
> based on the nerve ending studies of female vs. male populations) of
> the test population?
Myles,
I'm busy preparing a study, based on literature
available to me including both controlled listening
tests of audio components (JAES) and psychoacoustic
experiments (JAES and JASA). You might be interested
and read it in the audio asylum once it's finished.
Most of your questions will be answered there.
Attempt to answer some questions already :
1.How many people ?
In the list presented in
www.oakland.edu/djcarlst/abx_peri.htm there are some
papers from medical journals who might give the
answer. As far as listening tests and psychoacoustic
experiments are concerned, the greatest number of
participants I found sofar was 42 (loudspeaker
testing by Toole). In 1938, 1940 Fletcher conducted
tests on 500,000 visitors of the World Fair for
threshold of hearing (amplitude vs frequency), for
all ages, sexes and ethnic origins.
2.different hearing abilities :
these are tested before conducting a listening test
or psychoacoustic experiment. I don't think that
ethnic origin makes a difference since the sense of
hearing did not undergo further evolution since, say,
we made the step to homo sapiens. And even if, the
skills to be developped were the same everywhere,
based on one necessity : survival.
3.How were the tests carried out? Headphones ?
Headphones and speakers.
4. age :
highest age found sofar was 28 years. People older
than that start to "suffer" from hearing loss, which
makes them no longer useful fur meaningful listening
tests.
5. occupation :
musicians, audio engineers, hobbyists to name a few.
In listening tests of audio components special
attention may be paid to those having "golden ears".
6. female vs. male populations :
hearing loss due to age (+ occpupational exposure to
noise) is in general greater for men than for women.
But seen the age of subjects selected for tests, such
hearing loss is generally not a problem, and hearing
is tested beforehand anyway.
7. your hint to drug testing :
since for controlled listening tests and
psychoacoustical experiments otologically normal
subjects are selected, findings and thresholds
established by those are valid for 100% of the
remaining population (within the framework of
standards of what is otologically normal), because
within such framework any other subject is,
otologically seen, not normal, his level of
perception thus being compromised or decreased. The
small percentage of people having more than average
hearing capacities can be neglected.
Klaus
jj, curmudgeon and tiring philalethist
<KlausRa...@compuserve.com> wrote:
>In the list presented in
>www.oakland.edu/djcarlst/abx_peri.htm there are some
>papers from medical journals who might give the
>answer. As far as listening tests and psychoacoustic
>experiments are concerned, the greatest number of
>participants I found sofar was 42 (loudspeaker
>testing by Toole). In 1938, 1940 Fletcher conducted
>tests on 500,000 visitors of the World Fair for
>threshold of hearing (amplitude vs frequency), for
>all ages, sexes and ethnic origins.
You ought to also check out the various MPEG-Audio
competition and verification tests.
>since for controlled listening tests and
>psychoacoustical experiments otologically normal
>subjects are selected, findings and thresholds
>established by those are valid for 100% of the
>remaining population (within the framework of
>standards of what is otologically normal),
No, Klaus that's not right. In fact, for some of the studies
"otologically normal" is not a requirement, and for the simple
reason that training also plays a large part in how people
distinguish between stimulii in most kinds of DBT.
In fact, for instance, in some of the later MPEG studies,
the subjects were selected on the basis of their previous
performance, withOUT otological testing, and were verified
on the basis of their performace in the test setting.
And this is certainly not a wrong way to do it.
KlausRa...@compuserve.com
j...@research.att.com (jj, curmudgeon and tiring
philalethist) wrote:
>You ought to also check out the various MPEG-Audio
>competition and verification tests.
What I have direct access to is JAES, JASA, IEEE
transactions. I have found one of those ISO/MPEG
subjective tests on multichannel systems.
>No, Klaus that's not right. In fact, for some of
>the studies "otologically normal" is not a
>requirement, and for the simple reason that training
>also plays a large part in how people distinguish
>between stimulii in most kinds of DBT.
Most of the JASA psychoacoustic studies I have seen
do hearing threshold testing. Training is rarely
mentioned. Training does effectively increase
perception for stimuli, but I don't think that this
can overcome permanent hearing loss (age, noise
exposure, accidents).
For audio listening tests one of the authors stated
that one trained listener outweighs eight untrained.
> In fact, for instance, in some of the later MPEG
>studies, the subjects were selected on the basis of
>their previous performance, withOUT otological
>testing, and were verified on the basis of their
>performace in the test setting.>
> And this is certainly not a wrong way to do it.
Verification by test performance during audio
listening tests is done by some testers, although not
systematically. Some also perform hearing tests
before. There is, alas, not much technical literature
available about controlled listening tests. The tests
such as ABX or DBT as published in audio magazines
seem to suffer from one major drawback : not enough
listeners and not enough trials to obtain a sound
basis for meaningful statistics, the results thus
being questionable. This obviously applies for any
listening test as commonly published in those mags.
But that's a different story. It's called the great
debate.
jj, curmudgeon and tiring philalethist
<KlausRa...@compuserve.com> wrote:
>In article <8di93p$7vl$1...@news.aud.alcatel.com>,
> j...@research.att.com (jj, curmudgeon and tiring
>philalethist) wrote:
>>You ought to also check out the various MPEG-Audio
>>competition and verification tests.
>What I have direct access to is JAES, JASA, IEEE
>transactions. I have found one of those ISO/MPEG
>subjective tests on multichannel systems.
There is a very good AES report on a bunch of things in the
March 1998 JAES, too.
>Most of the JASA psychoacoustic studies I have seen
>do hearing threshold testing.
If you're testing basic attributes it makes sense. It
is much less relevant to testing using audio signals
and audio processing methods, of course, barring extensive
impairment, but that shows up in the subject performance
during test screening, as wlel.
>Training is rarely
>mentioned. Training does effectively increase
>perception for stimuli, but I don't think that this
>can overcome permanent hearing loss (age, noise
>exposure, accidents).
For listening to complex stimulii like music, it
is clearly and obviously the case that extensive
training is a necessity, at least for studying things
like coding artifacts for good perceptual coders.
This is not testing at absolute threshold limits
for the most part. For some, "recruitment" makes
them more, instead of less, sensitive to some
impairments, btw.
Obviously, one can not do better than the limits of
hearing, but they are barely relevant to testing
things like perceptual coders.
>For audio listening tests one of the authors stated
>that one trained listener outweighs eight untrained.
Or worse, even.
>Verification by test performance during audio
>listening tests is done by some testers, although not
>systematically.
Some do it systematically. Not sure what you're implying
here.
>The tests
>such as ABX or DBT as published in audio magazines
>seem to suffer from one major drawback : not enough
>listeners and not enough trials to obtain a sound
>basis for meaningful statistics, the results thus
>being questionable.
Well, there are other problems beyond that, editors are loath
to allow writers to include essential information on
methods, etc. Yes, I'm sure about this one. As a result
much of the discussion falls on the editor's floor.
There is also some confusion about the rigor necessary to
test a given subject's assertion of "obvious to a
deaf person" vs. that necessary to test "likely inaudible
to the general population".
Arny Krueger
news:8dkmfi$hkr$1...@news.aud.alcatel.com...
> In article <8di93p$7vl$1...@news.aud.alcatel.com>,
> j...@research.att.com (jj, curmudgeon and tiring
> philalethist) wrote:
>
> >You ought to also check out the various MPEG-Audio
> >competition and verification tests.
> What I have direct access to is JAES, JASA, IEEE
> transactions. I have found one of those ISO/MPEG
> subjective tests on multichannel systems.
A summary of the MPEG tests were written up in the JAES around March,
1999.
You can also download the more complete version of the tests from
http://www.cselt.it/mpeg/public/AAC_results.zip .
There is also a loose sort of testing of hearing capabilities that
has been ongoing at www.pcabx.com. This site provides a Windows
program that is a double blind comparator and a fair number of test
files to do varioius kinds of comparisons with. The current rate of
downloads of the comparator is about 10 per day, and the site has
been in operation for about 7 months. Careful records were not kept,
but its possible that several thousand ABX comparators have been
downloaded from this site and used in the past 7 months.
There has been exactly one report of reliable detection of an audible
difference between musical program material that was brick-wall
filtered at 18 KHz. There have been no reports of relaible detection
of an audible difference between musical program material that was
bit-truncated at 15 bits. There have been a great number of reports
of hearing audible differences between MP3 files coded at 128 KHz
versus the original files. There have been a great number of reports
of hearing audible differences due to recording and playback using
what is probably the most popular PC sound card on the market today.
There have been no reports of audible differences due to recording
and playback with a high end MD recorder.