A perrennial favorite here on rahe, and as it turns out,
discussed in this rather interesting article by Ian Dennis
and the late Julian Dunn

"The Numerically-IDentical CD Mystery: A Study in Perception versus Measurement"

The 'mystery' in question is the music biz/audiophile folklore that
bit-perfect copies can sound different -- specifically, pre-masters
versus the final CD product.

In esssence, the authors tested this perceptually via
blind comparison (by 'mass' listeners and by 'golden ears')
and via measurements of carefully-selected and prepared
test tracks.

The results are rather amusing, and speak to, in addition
to the titular matter:

- the effects of jitter (something Mr. Dunn in particular
had written about a lot before this)
- the ability of test subjects to follow directions,
- the validity of the 'golden ear' concept,
- one-box vs. two-box CD player designs
- the utility of controls in comparisons -- note
especially the results for the comparison of
same to same

Required reading, I think , for RAHEistas

http://www.prismsound.com/downloads/cdinvest.pdf





--

-S.

"They've got God on their side. All we've got is science and reason."
-- Dawn Hulsey, Talent Director

"Steven Sullivan" wrote in message
...
A perrennial favorite here on rahe, and as it turns out,
discussed in this rather interesting article by Ian Dennis
and the late Julian Dunn

"The Numerically-IDentical CD Mystery: A Study in Perception versus
Measurement"

The 'mystery' in question is the music biz/audiophile folklore that
bit-perfect copies can sound different -- specifically, pre-masters
versus the final CD product.

In esssence, the authors tested this perceptually via
blind comparison (by 'mass' listeners and by 'golden ears')
and via measurements of carefully-selected and prepared
test tracks.

The results are rather amusing, and speak to, in addition
to the titular matter:

- the effects of jitter (something Mr. Dunn in particular
had written about a lot before this)
- the ability of test subjects to follow directions,
- the validity of the 'golden ear' concept,
- one-box vs. two-box CD player designs
- the utility of controls in comparisons -- note
especially the results for the comparison of
same to same

Required reading, I think , for RAHEistas

http://www.prismsound.com/downloads/cdinvest.pdf


All I can say is they never would have held a job as market research test
designer at any company I worked for.

This test design is so bad in terms of what the evaluators are asked to do,
that once they found out the evaluators couldn't handle the instructions,
they should have thrown out the existing test and started over. There is so
much noise in this test (poor design, too few degrees of freedom) that
gaining meaningful results is almost doomed from the start.

Harry Lavo wrote:
"Steven Sullivan" wrote in message
...
A perrennial favorite here on rahe, and as it turns out,
discussed in this rather interesting article by Ian Dennis
and the late Julian Dunn

"The Numerically-IDentical CD Mystery: A Study in Perception versus
Measurement"

The 'mystery' in question is the music biz/audiophile folklore that
bit-perfect copies can sound different -- specifically, pre-masters
versus the final CD product.

In esssence, the authors tested this perceptually via
blind comparison (by 'mass' listeners and by 'golden ears')
and via measurements of carefully-selected and prepared
test tracks.

The results are rather amusing, and speak to, in addition
to the titular matter:

- the effects of jitter (something Mr. Dunn in particular
had written about a lot before this)
- the ability of test subjects to follow directions,
- the validity of the 'golden ear' concept,
- one-box vs. two-box CD player designs
- the utility of controls in comparisons -- note
especially the results for the comparison of
same to same

Required reading, I think , for RAHEistas

http://www.prismsound.com/downloads/cdinvest.pdf


All I can say is they never would have held a job as market research test
designer at any company I worked for.

This test design is so bad in terms of what the evaluators are asked to do,
that once they found out the evaluators couldn't handle the instructions,
they should have thrown out the existing test and started over. There is so
much noise in this test (poor design, too few degrees of freedom) that
gaining meaningful results is almost doomed from the start.

I'd say the worst part of the perceptual test design -- the measurement portion
strikes me as *extremely* methodical and thorough, though I'm no engineer --
was that it allowed too *much* freedom to the participants (and yes, I know 'degrees
of freedom' doesn't refer to that sort of freedom).

But what about the smaller test of 'golden eared' subjects, which
appears to have been carried out with more oversight than the
'mail in' part of the test?

And even if the instructions were inconsistently followed,
what are the chances of a result where the same-to-same
comparison produced among the HIGHEST 'difference' scores?

Of course, if you give no credence to these results because of
sloppiness or poor design, surely you have to maintain at least the
same level of skepticism towards the widespread anecdotal reports of
audible pressing differences that spawned the project.


--

-S.

"They've got God on their side. All we've got is science and reason."
-- Dawn Hulsey, Talent Director

On 22 Jan 2004 18:04:17 GMT, Steven Sullivan wrote:

A perrennial favorite here on rahe, and as it turns out,
discussed in this rather interesting article by Ian Dennis
and the late Julian Dunn

"The Numerically-IDentical CD Mystery: A Study in Perception versus Measurement"

The 'mystery' in question is the music biz/audiophile folklore that
bit-perfect copies can sound different -- specifically, pre-masters
versus the final CD product.
[snip]
http://www.prismsound.com/downloads/cdinvest.pdf

Somewhere in the nineties some artists, pop musicians and classical
musicians alike, began to complain that the final cd would sound
different, even worse (more compressed, more sharp) than the master
cdrom/tape they had approved of.

For a time nobody would believe them. But in the end people started to
investigate likely causes of the problem. It turned out that a
probable cause could be that SOMETHING went wrong at the pressing
plant. It turned out that some plants seemed to produce better cds
than others.

Is it really true that the pressing makes a difference for cds?

In a hearing test it is very difficult to come to conclusions,
because, as has been said ample times before,
(a) our hearing system is easily influenced by non-auditory
information
(b) our hearing system DEPENDS for good functioning on non-auditory
information.

So it is very easy to lead a test person to one or the other
direction. Furthermore, this moment the listener may concentrate on
this aspect of the sound/music, and another moment he may concentrate
on quite another aspect of the sound, which leads him to hear
different things at different times for the same sound source.

We all know this. If you listen to Beethoven's opus 131 for the 20th
time, you hear different things than the first time. So what you hear
changes, while the audio data remain the same.

This is the fundamental problem with hearing tests, whether blind or
sighted or done in one day or over many days.

With digital audio the problem is aggravated by the fact that what may
be audible with one piece of equipment, may not be audible with
another piece of gear. Generally the naive idea is that the gear that
makes for most audible differences is the "most sensitive" and hence
the "best". With analogue gear this may be true (turntables), but in
digital audio this idea is more times false then true.

A good example is the audibility of differences between digital
interlinks in two box cd systems. A box with not-so-good jitter
rejection (e.g. too simple PLL) depends very much on the quality of
the interlink. Slightly off 75 ohm and the signal starts to bounce and
whatever, resulting in audible jitter. So the simple system needs a
very good interlink.

A box with superb jitter rejection however, will not suffer that much
from a less than perfect interlink. Therefore, differences that are
audible with the simple box are inaudible with the very good box. So
the good piece of gear is "less sensitive".

As far as things have been tested, the same holds true of cd players
and different pressings. The better the cd player, the less
influential is the pressing. The worse the player is, the more it
depends on good pressings.

----------------

In 1994 Bob Katz wrote an article on the supposedly audible effects of
bad pressings in Stereophile:

http://www.stereophile.com/printarchives.cgi?55

In his book "Mastering Audio" (Focal Press 2002, isbn 0-240-80545-3),
a must-read for everybody interested in audio and how it sounds and
what makes it sound, he writes in his excellent chapter on jitter the
following on page 237:

"As I said, there is no jitter on a storage medium, but there is some
(controversial) evidence that CDs cut at high speeds sound inferior to
CDs cut at low speeds and that CDs cut with a jittery clock sound
worse than those cut with a clean clock."
In a note he adds:
"We theorize that irregular pit spacing or inadequate pit depth on the
CDs themselves is affecting the player's servo mechanism. The servo
mechanism and sample clock share a common power supply, so with poor
power supply bypass in the player, simple power or ground leakage may
affect the stability of the clock. It doesn't take much leakage to
change a few picoseconds."

Furthermore, back to page 237, Bob states:
"It only takes a few picoseconds to make an audible difference."

We may presume that the better the power supply of the cd player is,
the less will this detrimental effect of the servo be audible.

----------------

In the past I have written about the audibility of the display in my
TEAC vrds 8 cd player. Light on would add some high freqs to the
sound, most notable in some lute music, not notable in some symphonic
music. Generally I preferred the sound with the lights ON.

This player does not have the best of power supplies. Since then I
have inserted new clocks, together with a completely separate power
supply, that is, even a transformer of their own. It turned out that
soundwise this extra transformer was very important.

Also, it has become extremely difficult, or should I say generally
impossible to hear any audible influence of the display. This
corroborates the theory of Bob Katz: the display lights would
influence the power supply and the print board and through these the
clocks, leading to jitter. But now, as the clocks are powered
completely independent from the main supply, the display does no
longer influence the clocks.

-----------

A friend of mine has been working in a record shop for years and he
claims that consistently American pressings of a CD sound better than
German pressings. If possible, he always buys his CDs over the
internet, to get an American pressing. He says that German pressings
sound sharp, hollow, etc.

I cannot corroborate his story as I have not compared on a regular
basis American to Geman pressings.

----------------

Many people report that if they burn their CD copies at home, they
sound worse if burned at high speed and better if burned at low speed
(4 times is low speed, 20 to 40 times is high speed).

Again, with a good player this may be less audible or even inaudible.

---------------

Regarding the article by Dennis & Dunn, it must be said that they try
to measure a lot of things together at the same time, usually not a
good idea if you want to achieve valuable results.

Ernesto.

"You don't have to learn science if you don't feel
like it. So you can forget the whole business if
it is too much mental strain, which it usually is."

Richard Feynman

Steven Sullivan wrote:

Harry Lavo wrote:
"Steven Sullivan" wrote in message
...
A perrennial favorite here on rahe, and as it turns out,
discussed in this rather interesting article by Ian Dennis
and the late Julian Dunn

"The Numerically-IDentical CD Mystery: A Study in Perception versus
Measurement"

The 'mystery' in question is the music biz/audiophile folklore that
bit-perfect copies can sound different -- specifically, pre-masters
versus the final CD product.

In esssence, the authors tested this perceptually via
blind comparison (by 'mass' listeners and by 'golden ears')
and via measurements of carefully-selected and prepared
test tracks.

The results are rather amusing, and speak to, in addition
to the titular matter:

- the effects of jitter (something Mr. Dunn in particular
had written about a lot before this)
- the ability of test subjects to follow directions,
- the validity of the 'golden ear' concept,
- one-box vs. two-box CD player designs
- the utility of controls in comparisons -- note
especially the results for the comparison of
same to same

Required reading, I think , for RAHEistas

http://www.prismsound.com/downloads/cdinvest.pdf


All I can say is they never would have held a job as market research test
designer at any company I worked for.

This test design is so bad in terms of what the evaluators are asked to do,
that once they found out the evaluators couldn't handle the instructions,
they should have thrown out the existing test and started over. There is so
much noise in this test (poor design, too few degrees of freedom) that
gaining meaningful results is almost doomed from the start.

I'd say the worst part of the perceptual test design -- the measurement portion
strikes me as *extremely* methodical and thorough, though I'm no engineer --
was that it allowed too *much* freedom to the participants (and yes, I know 'degrees
of freedom' doesn't refer to that sort of freedom).

I find the measurements where they revealed power-supply related jitter
and AM in a single-box player very interesting. I just wish someone
could include that kind of measurements in the reviews of disc players.
One plot is worth more than 1,000 words of purple prose to me.


But what about the smaller test of 'golden eared' subjects, which
appears to have been carried out with more oversight than the
'mail in' part of the test?

And even if the instructions were inconsistently followed,
what are the chances of a result where the same-to-same
comparison produced among the HIGHEST 'difference' scores?

Of course, if you give no credence to these results because of
sloppiness or poor design, surely you have to maintain at least the
same level of skepticism towards the widespread anecdotal reports of
audible pressing differences that spawned the project.

"Steven Sullivan" wrote in message
...
Harry Lavo wrote:
"Steven Sullivan" wrote in message
...
A perrennial favorite here on rahe, and as it turns out,
discussed in this rather interesting article by Ian Dennis
and the late Julian Dunn

"The Numerically-IDentical CD Mystery: A Study in Perception versus
Measurement"

The 'mystery' in question is the music biz/audiophile folklore that
bit-perfect copies can sound different -- specifically, pre-masters
versus the final CD product.

In esssence, the authors tested this perceptually via
blind comparison (by 'mass' listeners and by 'golden ears')
and via measurements of carefully-selected and prepared
test tracks.

The results are rather amusing, and speak to, in addition
to the titular matter:

- the effects of jitter (something Mr. Dunn in particular
had written about a lot before this)
- the ability of test subjects to follow directions,
- the validity of the 'golden ear' concept,
- one-box vs. two-box CD player designs
- the utility of controls in comparisons -- note
especially the results for the comparison of
same to same

Required reading, I think , for RAHEistas

http://www.prismsound.com/downloads/cdinvest.pdf


All I can say is they never would have held a job as market research
test
designer at any company I worked for.

This test design is so bad in terms of what the evaluators are asked to
do,
that once they found out the evaluators couldn't handle the
instructions,
they should have thrown out the existing test and started over. There
is so
much noise in this test (poor design, too few degrees of freedom) that
gaining meaningful results is almost doomed from the start.

I'd say the worst part of the perceptual test design -- the measurement
portion
strikes me as *extremely* methodical and thorough, though I'm no
engineer --
was that it allowed too *much* freedom to the participants (and yes, I
know 'degrees
of freedom' doesn't refer to that sort of freedom).

But what about the smaller test of 'golden eared' subjects, which
appears to have been carried out with more oversight than the
'mail in' part of the test?

And even if the instructions were inconsistently followed,
what are the chances of a result where the same-to-same
comparison produced among the HIGHEST 'difference' scores?

Of course, if you give no credence to these results because of
sloppiness or poor design, surely you have to maintain at least the
same level of skepticism towards the widespread anecdotal reports of
audible pressing differences that spawned the project.


The fact that the "no difference" disk ended up "high" is prima-facia
evidence that the test was sloppy. If all biases in the test design were
properly nulled and the degrees of freedom adequate, the result would have
been "average" due to random selection in all disks were the same, or lower,
if some disks truly were audibly different and the test design allowed this
to be detected.

As far as the issue itself, I am an agnostic. I have no first hand
experience or theoretical knowledge or even interest in the subject to have
an opinion.