[Arny Krueger]

"Randy Yates" wrote in message

"Arny Krueger" writes:
[...]
One the last things in audio to go wrong was so-called
high resolution audio for distribution to consumers.
Serious money was put into DVD-A and SACD, and they both
failed to gather moementum in the mainstream
marketplace. In contrast, there was a concurrent effort
called high definition video, and there seems to be a
great number of indications that this will become the
next mainstream technology.

I think this is as much an indication that the business
model is flawed as anything else.

The business model that ties a hoard of income to
royalties for the
basic technology is always going to encourage format wars.

In the case of SACD, you probably nailed it, Randy.

SACD involved a fair amount of technology that was probably both patentable,
and also patented. At one point, Sony seemed to be trying to talk the world
into going DSD from someplace inside the microphone case, to someplace
within the speaker case, and everyplace in-between.

If the audio world had bitten into the DVD hype, Sony and Philips stood to
make a ton of money, given reasonable competence on their part.

Thus there was a clear profit motive behind all the absolutely
anti-scientific claims that DVD sounded better than PCM.

In contrast, DVD-A seemed to involve virtually nothing that was not already
prior art, just the clock speed goosed somewhat.

[Chris Hornbeck]

On Mon, 17 Dec 2007 08:57:18 +0100, "Peter Larsen"
wrote:

Perfect and true, but incomplete.

It is not true, what is true is that the group of testpersons could not hear
a difference, you can only make probability predictions based on such data,
a very high probability may be reachable but not absolute knowledge. Always
read the fine print that goes with the test .....

Properly blinded testing can remove false positives (a classical
issue in the audio field, fersure) but leave open the issue of
false negatives.

My personal take is that properly blinded testing is a useful
winnowing in intermediately difficult judgements. In other
situations, it falls down.

In a situation where the slippery objective Truth, whatever that
means, is below the test's rigor, it fails. (False negative).

And in a situation where a simple difference is less important to
the goal of the testing than a ranking, it fails. (False ranking).
This will be controversial, so maybe shouldn't be introduced here.
But it's introduced to remind us all that nothing is perfect and
that *all* decisions are provisional, all hardware and all
observations are imperfect, all judgements inherently biased:

"Life is short
Art is long
Experiment treacherous
Judgement difficult."
-Hippocrates



Much thanks, as always,

Chris Hornbeck

[Arny Krueger]

"Chris Hornbeck" wrote in
message

Properly blinded testing can remove false positives (a
classical issue in the audio field, fersure) but leave
open the issue of false negatives.

The logical failure here is the lack of a balanced discussion of the
equally-serious or even more serious problem of false negatives during
sighted evaluations.

One of the ironies of life is the fact that I built the first ABX Comparator
with the main goal of avoiding false negatives during sighted evaluations. I
was so optimistic about the possibility of nearly universal positive results
relating to the audio issues that I was interested in at the time such as
differences between power amplifiers, that I was totally surprised when so
many tests came out negative.

My reasoning was that a blind test tends to decouple the listener from his
biases. In a blind test, the listener doesn't know what kind of response to
give in each trial, in order to produce the overall result that he expects
and/or desires. If he intentionally randomizes his responses, then he is
singled out as being an insensitive listener.

There are many causes of false negatives during sighted evaluations. Here
are just a few:

(1) The listener's biases were in favor of a negative outcome but the
correct answer is a positive outcome. Since a sighted evaluation has
inadequate bias controls, the listener's report of a negative result could
be due to his biases, not based on what he heard.

(2) Many sighted evaluations are either/or tests. The user responds in
accordance with his biases which results in an outcome that he favors and/or
that agrees with his biases. However, the correct answer is the other
outcome. His response is a false negative.

(3) In many sighted evaluations, the presumed correct answer is merely
determined by someone's biases. The person who has decided what the right
answer is, could be exactly wrong. Therefore, all the people that agree with
him, would then be reporting a false negative.

(4) A listener may feel that he is biased towards a certain outcome, and in
an effort to appear to be unbiased, overcompensate and report a false
negative for the outcome that he thinks he is biased toward.

[Steven Sullivan]

Chris Hornbeck wrote:
On Sun, 16 Dec 2007 19:18:29 +0000 (UTC), Steven Sullivan
wrote:

Chris Hornbeck wrote:

Figured as much.

So you don't do much recording, do you?

No, but my standards of proof are set by science. Do much of that?

I try to. And I believe that my standards of proof are set
pretty high. We will probably differ very little here, but
ya never can tell what any particular person will consider
"science" and another consider "religion".

Once a shouting match begins, rationality goes out the window.

A good reason to utterly avoid places like rec.audio.opinion and
audioasylum.com

For an island of audio sanity I tend to rely on hydrogenaudio.org.
Do you post there?


___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Steven Sullivan]

Arny Krueger wrote:
"Steven Sullivan" wrote in message

Chris Hornbeck wrote:
On Fri, 14 Dec 2007 17:51:30 +0000 (UTC), Steven Sullivan
wrote:

In rec.audio.tech Peter Larsen
wrote:
Steven Sullivan wrote:

In rec.audio.tech Peter Larsen
wrote:
geoff wrote:

Why 96/16 rather than 44k1/24 ? I don't follow
that logic.

Because the treble sounds cleaner with better
inter-transient silence, and that really matters
with decayed audio, it gets less splatty.

Do tell. Your proof of this is....?

I stated an opinion. I do not waste time proving
recording choices, I make them based on what sounds
best.

Figured as much.

So you don't do much recording, do you?

No, but my standards of proof are set by science. Do much
of that? If you do , you'll understand that claims
should be independently verifiable. IF what you guys
hear is real, it should be readily verified in a
controlled listening test, no?

Note that two recent AES papers have come out relating to so-called high
resolution audio. In both cases it looks to me like the experimenters made a
heck of a try, but both sets of test results showed essentially random
guessing. You know, something like 50% or less of the listener choices
favored so-called high definition audio as compared to CD-format audio.

But one possibly significant difference is that Meeyer & Moran (2007) was in JAES, thus
peer-reviewed, while the Woszczyk & Usher (200& AES conference paper, was not. Assuming those
are the two you're referring to.


___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Steven Sullivan]

Chris Hornbeck wrote:
On Mon, 17 Dec 2007 19:01:46 -0500, Randy Yates
wrote:

Chris Hornbeck writes:
[...]
It would be interesting to read your equally well put description
of the contrapositive position.

Which logical statement are you requesting the contrapositive
of?

If something is audible, then it can be heard in a double-blind test.
^
v
If something can't be heard in a double-blind test, then it isn't audible.

That one. (I'm sorry that this was unclear. Arny and I have
danced this dance before, and can shortcut a lot. I greatly
admire his work, but function as a gadfly in some areas of
mutual interest.)

Science deals in probabilities. You know this, I know this, Arny knows this, audiophiles who
use the 'never say never' argument may know this. But the latter often twist its meaning to
imply that the negative DBT means NOTHING, when in fact it's another data point -- another
sample of the population -- showing lack of support for a hypothesis of audible difference.
Audiophiles tend to think that the possibility of someone else hearing a difference remains
the same, no matter how many times 'no difference' is achieved. That's certainly not how
science works. Science doesn't ignore current data and instead hold all likelihoods as equal.


___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Steven Sullivan]

Chris Hornbeck wrote:
On Mon, 17 Dec 2007 08:57:18 +0100, "Peter Larsen"
wrote:

Perfect and true, but incomplete.

It is not true, what is true is that the group of testpersons could not hear
a difference, you can only make probability predictions based on such data,
a very high probability may be reachable but not absolute knowledge. Always
read the fine print that goes with the test .....

Properly blinded testing can remove false positives (a classical
issue in the audio field, fersure) but leave open the issue of
false negatives.

Protocols can account for both Type I and Type II errors.

My personal take is that properly blinded testing is a useful
winnowing in intermediately difficult judgements. In other
situations, it falls down.

In a situation where the slippery objective Truth, whatever that
means, is below the test's rigor, it fails. (False negative).

Please give a real-world example of such 'slippery objective Truth'.

And in a situation where a simple difference is less important to
the goal of the testing than a ranking, it fails. (False ranking).

Proper 'ranking' tests are done blind, too.

This will be controversial, so maybe shouldn't be introduced here.
But it's introduced to remind us all that nothing is perfect and
that *all* decisions are provisional, all hardware and all
observations are imperfect, all judgements inherently biased:

It's introduced without evidence, and that's what rankles.


___
-S
"As human beings, we understand the world through simile, analogy,
metaphor, narrative and, sometimes, claymation." - B. Mason

[Arny Krueger]

"Steven Sullivan" wrote in message

Arny Krueger wrote:
"Steven Sullivan" wrote in message

Chris Hornbeck wrote:
On Fri, 14 Dec 2007 17:51:30 +0000 (UTC), Steven
Sullivan wrote:

In rec.audio.tech Peter Larsen
wrote:
Steven Sullivan wrote:

In rec.audio.tech Peter Larsen
wrote:
geoff wrote:

Why 96/16 rather than 44k1/24 ? I don't follow
that logic.

Because the treble sounds cleaner with better
inter-transient silence, and that really matters
with decayed audio, it gets less splatty.

Do tell. Your proof of this is....?

I stated an opinion. I do not waste time proving
recording choices, I make them based on what sounds
best.

Figured as much.

So you don't do much recording, do you?

No, but my standards of proof are set by science. Do
much of that? If you do , you'll understand that claims
should be independently verifiable. IF what you guys
hear is real, it should be readily verified in a
controlled listening test, no?

Note that two recent AES papers have come out relating
to so-called high resolution audio. In both cases it
looks to me like the experimenters made a heck of a try,
but both sets of test results showed essentially random
guessing. You know, something like 50% or less of the
listener choices favored so-called high definition audio
as compared to CD-format audio.

But one possibly significant difference is that Meeyer &
Moran (2007) was in JAES, thus peer-reviewed, while the
Woszczyk & Usher (200& AES conference paper, was not.

Right. So the JAES article, being peer-reviewed, is the far more
authoritative of the two.

The conference paper showed a lot of experimental vigor, but taxes one's
ability to suspend disbelief, given that their results were so close to 50%
inconsistent results, or worse.

We need a professor Bernstein (sp?) to elucidate the problems with its
statistical analyis.

Assuming those are the two you're referring to.

Right.

[Chris Hornbeck]

On Tue, 18 Dec 2007 20:45:12 +0000 (UTC), Steven Sullivan
wrote:

Once a shouting match begins, rationality goes out the window.

A good reason to utterly avoid places like rec.audio.opinion and
audioasylum.com

For an island of audio sanity I tend to rely on hydrogenaudio.org.
Do you post there?

No, only in rec.audio.pro these days. My hobby newsgroups have
gotten contaminated; and I'm only a tourist here. But from the
generally high tone seen here lately, I'll be reading here
alot.

Thanks, as always,

Chris Hornbeck

[Chris Hornbeck]

On Tue, 18 Dec 2007 20:57:36 +0000 (UTC), Steven Sullivan
wrote:

In a situation where the slippery objective Truth, whatever that
means, is below the test's rigor, it fails. (False negative).

Please give a real-world example of such 'slippery objective Truth'.

All "capital T" truth is slippery. To my way of thinking, our
incredible ignorance of the listening process/machinery should
always be forefront when belief is attempted.

But hey, I'm a pretty conservative guy. Makes me vewy, vewy
agenda-averse and largely obvious-conclusion-averse.

Thanks, as always,

Chris Hornbeck

[Chris Hornbeck]

On Tue, 18 Dec 2007 13:22:39 -0500, "Arny Krueger"
wrote:

My reasoning was that a blind test tends to decouple the listener from his
biases. In a blind test, the listener doesn't know what kind of response to
give in each trial, in order to produce the overall result that he expects
and/or desires. If he intentionally randomizes his responses, then he is
singled out as being an insensitive listener.

My light-bulb-on moment was in the mid-1970's at the retail audio
store that employed me (and does for the fourth time today!). I
boldly claimed that I could tell a McIntosh from a Phase Linear
amplifier in the showroom by listening alone.

Couldn't.


There are many causes of false negatives during sighted evaluations. Here
are just a few:

list snipped for bandwidth

False negatives during blinded testing are the much more pertinent
and useful issue.

Much thanks, as always,

Chris Hornbeck

[Chris Hornbeck]

On Tue, 18 Dec 2007 20:57:36 +0000 (UTC), Steven Sullivan
wrote:

and that's what rankles.

Rereading, I *finally* caught the pun. Arf!

Much thanks, as always,

Chris Hornbeck

[Arny Krueger]

"Chris Hornbeck" wrote in
message
On Tue, 18 Dec 2007 13:22:39 -0500, "Arny Krueger"
wrote:

My reasoning was that a blind test tends to decouple the
listener from his biases. In a blind test, the listener
doesn't know what kind of response to give in each
trial, in order to produce the overall result that he
expects and/or desires. If he intentionally randomizes
his responses, then he is singled out as being an
insensitive listener.

My light-bulb-on moment was in the mid-1970's at the
retail audio store that employed me (and does for the
fourth time today!). I boldly claimed that I could tell a
McIntosh from a Phase Linear amplifier in the showroom by
listening alone.

Couldn't.

Been there, done that.

There are many causes of false negatives during sighted
evaluations. Here are just a few:

list snipped for bandwidth

False negatives during blinded testing are the much more
pertinent and useful issue.

Why?

A lot more sighted evaluations are done than blind tests. Anything that
afflicts the most listening evaluations should be of greater interest.

[Chris Hornbeck]

On Wed, 19 Dec 2007 07:53:37 -0500, "Arny Krueger"
wrote:

False negatives during blinded testing are the much more
pertinent and useful issue.

Why?

A lot more sighted evaluations are done than blind tests. Anything that
afflicts the most listening evaluations should be of greater interest.

Because unblinded evaluations are pretty much useless for
anything of real interest. It's a dead horse.

I'm interested in moving forward, technically if not spiritually
(Arf), so discount unblinded evaluations, except on TV's.

And I make an exception in the case of very long term use
by professionally critical folks with no obvious agenda.
If such a person were to tell me something that conflicted
with a properly blinded difference eliminating test, I would
begin by studying the flaws of the test itself, looking, usually,
for false negatives.

We've done this before, so both our positions here are clear,
I hope. We're each fighting for the higher ground scientifically
from within our own personal religious framework. You say that your
religion is science; I say that my religion is science. It's the
story of the world; thus has it been, thus will it ever be.

Much thanks, as always,

Chris Hornbeck

[why bother]

"tony sayer" wrote in message
...
In article , Arny Krueger

This is a UK froop, the last poster we want in here is
Arny Krueger, whatever the subject, no matter the
thread evolution AK will eventually resort to his
cyberstalking of John Aitkinson, and Arny's ******** about
ABX and his highly dubious contentions...yawn.
If you're placing any credibilty in Arny's opnions or views
then go ask him in the froops he haunts, he's not wanted
here in the land of the sane and knowledgeable.

[John Corbett]

In article , "Arny Krueger"
wrote:

We need a professor Bernstein (sp?) to elucidate the problems with its
statistical analyis.

Be careful what you wish for. ;-)

Herman Burstein wasn't the only one who could understand the statistical
issues involved in listening tests. Statisticians do that sort of thing
all the time.
Now it's the end of the semester, and I have piles of final exams to grade.
After I've submitted final course grades, I'll post some comments on both
of the AES articles mentioned earlier in this thread.

[Randy Yates]

(John Corbett) writes:

In article , "Arny Krueger"
wrote:

We need a professor Bernstein (sp?) to elucidate the problems with its
statistical analyis.

Be careful what you wish for. ;-)

Herman Burstein wasn't the only one who could understand the statistical
issues involved in listening tests. Statisticians do that sort of thing
all the time.
Now it's the end of the semester, and I have piles of final exams to grade.
After I've submitted final course grades, I'll post some comments on both
of the AES articles mentioned earlier in this thread.

I look forward to it, John.
--
% Randy Yates % "Remember the good old 1980's, when
%% Fuquay-Varina, NC % things were so uncomplicated?"
%%% 919-577-9882 % 'Ticket To The Moon'
%%%% % *Time*, Electric Light Orchestra
http://www.digitalsignallabs.com

[Arny Krueger]

"Randy Yates" wrote in message

(John Corbett) writes:

In article
, "Arny
Krueger" wrote:

We need a professor Bernstein (sp?) to elucidate the
problems with its statistical analyis.

Be careful what you wish for. ;-)

Herman Burstein wasn't the only one who could understand
the statistical issues involved in listening tests.
Statisticians do that sort of thing all the time.
Now it's the end of the semester, and I have piles of
final exams to grade. After I've submitted final course
grades, I'll post some comments on both of the AES
articles mentioned earlier in this thread.

I look forward to it, John.

As will I.

[Adrian]

On Dec 11 2007, 10:37*am, "Arny Krueger" wrote:
"Richard Crowley" wrote in message







"Dave W." *wrote ...
*Adrian*wrote:
This past weekend I copied three albums. *The signal is
clean but not strong. *I have the gain on theUSBPhono
turned to the max.

If, as you say, the signal is clean, then as long as you
have it digitised there is no problem.

Lets review the bidding....

* *Low-output MC cartridge feeding an inexpensive RIAA
phono preamp designed for MC.
* *Gain on the preamp "turned to the max".
* *Signal is "clean but not strong"
Therefore, by definition, the captured signal is NOT
"clean" after amplifying it (plus the noise) to the
nominal level.

Agreed.

Of course,Adriancould decide that it is good enough
for his purposes, and that is fine. *But conventional
wisdom would suggest that the solution might be...
1) Use a conventional MM cartridge
2) Use a step-up transformer or pre-pre-amp for MC
3) Use a preamp designed for MC.

I'd vote for solution number 1, more specificially this cartridge:

http://www.amazon.com/Shure-M97xE-Hi...etic-Cartridge...

Cheapest way out and solves more problems.

Solution implemented. The results, to quote Pop Larkin are "Perfick",
or at least somewhat as close as we ever get to perfect in the world
of audio reproduction. :-)

Many thanks

Adrian