[Audio Empire]

On Wed, 13 Oct 2010 16:03:18 -0700, jwvm wrote
(in article ):

On Oct 11, 6:31=A0pm, Audio Empire wrote:



The question is simple enough. I've asked not for evidence, but rather fo=
r an
explanation. How do YOU know that none of these amp designers, men like
Nelson Pass, John Curl, Jim Di Paravancini, William Z. Johnson, et al, =
=A0NEVER
SAT DOWN AND ACTUALLY TRIED A RELIABLE EXPERIMENT?


Are we supposed to be impressed with this list of names? Building high-
quality amplifiers these days is not very difficult. Again, of course,
without proper unbiased evaluation, all sorts of claims are likely to
be made about their designs.




On the other side of the fence, we have people like you and Mr. Kruger wh=
o
seem to be content to take the word of more-or-less anonymous, but publis=
hed
DBTs that support your assertion that there is no difference. Never once =
I
have seen either of you assert that YOU have partaken of these tests (and=
if
I have misread that, I humbly apologize), known that they were properly
set-up and conducted, or that you have personal experience that tells you
that there is no difference between 16/44.1 and 24/96 or 24/192.

You are misconstruing things here. Arnie et al are not claiming there
is no perceptual difference. They are merely asking for properly
documented evidence that what you claim has a basis in fact.

I'm misconstruing NOTHING, my friend. Both Mr. Kruger and Mr. Pierce have
stated flat -out that there is no statistically meaningful perceptual
difference between 16/44.1 and 24/96 or 192 audio. I have outlined a number
of experiements that will show the difference WITHOUT needing an elaborate
DBT set up. The easiest is to take an LP that images very well (The Mercury
recording of Stravinsky's "Firebird" Ballet with Antal Dorati and the LSO is
a good example here) and quantize it with both 16/44.1 and 24/96 and then
listen to the LP - noting the soundstage and instrument locations in space,
then play the 16/44.1 copy of the record, and listen for same things. The
image specificity will be much less thre-dimensional and less palpable.
You'll be a lot less able to close your eyes and point to each instrument in
space. You will also notice that the soundstage will seem to collapse and
won't be a wide. Now, play the 24/96 of the same record and everything is
back where it should be. Try the same experiment with a record or analog
master tape of solo piano of perhaps a clavichord (which plays very softly).
notice the ambience and low-level detail on the analog recording, then listen
at both 16/44.1 and 24/96 (or higher) notice the truncated ambience and lack
of low level detail on the 16/44.1 recording and how it's back at the
24-bit/96KHz playback. These results are very unambiguous. The cues we are
talking about are either there are they aren't.

[Audio Empire]

On Thu, 14 Oct 2010 05:47:09 -0700, Dick Pierce wrote
(in article ):

Audio Empire wrote:
I'm misconstruing NOTHING, my friend. Both Mr. Kruger and Mr. Pierce have
stated flat -out that there is no statistically meaningful perceptual
difference between 16/44.1 and 24/96 or 192 audio.

Precisely when did I make such a statement, Mr. Empire?



I don't know. That is your assertion is it not? If I'm wrong and have
misattributed such an attitude to you, I humbly apologize.

[jwvm]

On Oct 13, 9:07=A0pm, Audio Empire wrote:
On Wed, 13 Oct 2010 16:03:18 -0700, jwvm wrote
(in article ):









On Oct 11, 6:31=3DA0pm, Audio Empire wrote:

The question is simple enough. I've asked not for evidence, but rather=
fo=3D
r an
explanation. How do YOU know that none of these amp designers, men lik=
e
Nelson Pass, John Curl, Jim Di Paravancini, William Z. Johnson, et al,=
=3D
=3DA0NEVER
SAT DOWN AND ACTUALLY TRIED A RELIABLE EXPERIMENT?

Are we supposed to be impressed with this list of names? Building high-
quality amplifiers these days is not very difficult. Again, of course,
without proper unbiased evaluation, all sorts of claims are likely to
be made about their designs.

On the other side of the fence, we have people like you and Mr. Kruger=
wh=3D
o
seem to be content to take the word of more-or-less anonymous, but pub=
lis=3D
hed
DBTs that support your assertion that there is no difference. Never on=
ce =3D
I
have seen either of you assert that YOU have partaken of these tests (=
and=3D
=A0if
I have misread that, I humbly apologize), known that they were properl=
y
set-up and conducted, or that you have personal experience that tells =
you
that there is no difference between 16/44.1 and 24/96 or 24/192.

You are misconstruing things here. Arnie et al are not claiming there
is no perceptual difference. They are merely asking for properly
documented evidence that what you claim has a basis in fact.

I'm misconstruing NOTHING, my friend. Both Mr. Kruger and Mr. Pierce have
stated flat -out that there is no statistically meaningful perceptual
difference between 16/44.1 and 24/96 or 192 audio. I have outlined a numb=
er
of experiements that will show the difference WITHOUT needing an elaborat=
e
DBT set up. The easiest is to take an LP that images very well (The Mercu=
ry
recording of Stravinsky's "Firebird" Ballet with Antal Dorati and the LSO=
is
a good example here) and quantize it with both 16/44.1 and 24/96 and then
listen to the LP - noting the soundstage and instrument locations in spac=
e,
then play the 16/44.1 copy of the record, and listen for same things. The
image specificity will be much less thre-dimensional and less palpable.
You'll be a lot less able to close your eyes and point to each instrument=
in
space. You will also notice that the soundstage will seem to collapse and
won't be a wide. Now, play the 24/96 of the same record and everything is
back where it should be. Try the same experiment with a record or analog
master tape of solo piano of perhaps a clavichord (which plays very softl=
y).
notice the ambience and low-level detail on the analog recording, then li=
sten
at both 16/44.1 and 24/96 (or higher) notice the truncated ambience and l=
ack
of low level detail on the 16/44.1 recording and how it's back at the
24-bit/96KHz playback. These results are very unambiguous. The cues we ar=
e
talking about are either there are they aren't.

Then how do you discount this AES paper that has proper controls and
every effort was made to eliminate bias:

http://www.aes.org/e-lib/browse.cfm?elib=3D14195

[Andrew Barss]

Audio Empire wrote:
: On Thu, 14 Oct 2010 06:41:37 -0700, Arny Krueger wrote
: (in article ):

: I said that I know of no good experiments with positive results for
: comparisons of that nature, and that I try to keep track of such things.
:
: I don't have any problems with the above paraphrase of my statements on this
: matter.
:
: I have outlined a number
: of experiements that will show the difference WITHOUT
: needing an elaborate DBT set up.

You need to seprate the words "elaborate" and "DBT" (double-blind test, or experiment).
Elabortion ones are used only when simpler ones can't be (I know, I am associate director of
a behavioral research laboratory).

But experiments or "tests" that are not double-blind are really, really, really, not ones you
want to be asssociated with, in any area of study. Period. DBTs are the ONLY way you have any certainty
that the results you get, e.g., "X sounds better than Y", consistently) are due to the factors you think
they are, like inherent differences in the audio signals from X and Y.

This is just basic science. Single-blind and non-blind tests are deficient.
The subject has to not know whether it's X or Y; the experimenter flipping between
X and Y has to not know this as well (hence the appeal of the sort of testing paradigm Arny helpd design,
in which a computer does this); and the person or machine coding the responses has to not know it as well.

So, EVEN IF you run the right number of trials, or the right number of subjects,
and you roperly randomize the assignment of stimuli to subjects, you STILL
*cannot know in principle* that there isn't influence on the outcome by the
bias of the experimenter. This is basic, undergraduate research methods.


: The image specificity will be much less
: thre-dimensional and less palpable.
:
: Why do the experiment when we have someone to tell us the results with such
: assuredness? ;-)

: The experiment is repeatable and the results are consistent. If you want to
: take my word for it, be my guest, but I really think that folks ought to find
: out these things for themselves.

No, actually, they should do that, and THEN do or read experiments which
show trustable results. The experiment will show one of two things:

(a) there is a consistent difference. And then you can publish the results!
(b) There is no evidence for a consistent difference. THEN you need to reinterpret
the "see it for yourself"/"I know it's true" data in an entirely different light.

-- Andy Barss

[Ed Seedhouse[_2_]]

On Oct 15, 8:05=A0am, Dick Pierce
wrote:
=A0 =A0The situation is becoming increasingly depressing.

Yes it is, though this is not, alas, limited to discussion of audio
topics.

Now if someone just says "I prefer the sound of vinyl" there should be
no gainsaying that. It's simply a fact that some persons do indeed
prefer the sound of vinyl. More power to them, and may they die at
ninety something, old and happy still listening to LP's.

And if they say "To me, vinyl sounds more like real music played in
real space" I have no quibble. I believe them. To them, in their
experience and opinion, vinyl sounds "more like real music played in
real space". Fine by me, live long and prosper. I have different
opinions, but why should they care about my opinions?

It's when they take the next step, the step that says "because I
prefer the sound of vinyl, therefore vinyl is better objectively" the
rationalist will step forward and say "hold on, just because you
prefer something over something else doesn't mean it is objectively
better, so where's your evidence?"

And that's where the arguments seem to start, and what they seem to me
to be all about. And, in my opinion, they could all be avoided if the
people who prefer vinyl over other media would simply keep their
opinions and express them without taking that extra step and claiming
that it is not merely a preference, but an objective fact.

I also, frankly, don't understand the objection some people have to
the suggestion that they might prefer something which is objectively
distorted to something that isn't. I mean, I would prefer to have a
self portrait of Vincent Van Gogh on my living room wall than to have
a colour photograph of him. The painting is objectively the more
distorted of the two, surely.

If you can prefer an objectively distorted work of art over an
objectively less distorted photograph, why would you object to
admitting that you prefer the more objectively distorted phonograph
record over the more objectively accurate CD?

[UnsteadyKen]

Arny Krueger said...

The only current vinyl renaissance that I know of being represented by
actual hardware involves sub-$200 playback systems which covers most of the
new LP player hardware purchases that I am aware of.

One purchase appears to have slipped in under your radar, Arny:-(

http://www.avidhifi.co.uk/medvedev.htm

--
Ken O'Meara
http://www.btinternet.com/~unsteadyken/

[H Davis]

"Audio Empire" wrote in message
...

snip

Massenburg then used a demonstration to drive his point home. He
electronically subtracted the MP3 compressed music from the original
24-bit/96 KHz recording and then played ONLY the difference signal which
was
comprised solely of the information lost by the compression.

"These are distortion levels of 15 � 20 percent! ", he said as he played
the
difference signal for all to hear. The distortion amazed everyone in
attendance because it was a grotesquely, but very recognizable version of
the
original recoding!

He went on to say that while AAC was clearly better than MP3, it still
generated 5% to 10% distortion.

snip

I have just one question. If indeed distortion levels of the magnitudes
quoted above are present, WHY ARE THEY NOT AUDIBLE? One can certainly hear
even one or two percent harmonic distortion.

[Audio Empire]

On Fri, 15 Oct 2010 11:26:36 -0700, UnsteadyKen wrote
(in article ):

Arny Krueger said...

The only current vinyl renaissance that I know of being represented by
actual hardware involves sub-$200 playback systems which covers most of the
new LP player hardware purchases that I am aware of.

One purchase appears to have slipped in under your radar, Arny:-(

http://www.avidhifi.co.uk/medvedev.htm



Wonder if Medvedev has an extensive Melodiya collection? I have a few from
the late seventies, early eighties (real ones, not the Capitol pressed Angel
issues) and they really sound good. 8^)

[jwvm]

On Oct 15, 2:38=A0pm, "H Davis" wrote:
"Audio Empire" wrote in message

...

snip

Massenburg then used a demonstration to drive his point home. He
electronically subtracted the MP3 compressed music from the original
24-bit/96 KHz recording and then played ONLY the difference signal whic=
h
was
comprised solely of the information lost by the compression.

"These are distortion levels of 15 20 percent! ", he said as he played
the
difference signal for all to hear. The distortion amazed everyone in
attendance because it was a grotesquely, but very recognizable version =
of
the
original recoding!

He went on to say that while AAC was clearly better than MP3, it still
generated 5% to 10% distortion.

snip

I have just one question. If indeed distortion levels of the magnitudes
quoted above are present, WHY ARE THEY NOT AUDIBLE? One can certainly hea=
r
even one or two percent harmonic distortion.

Because the distortion is primarily linear. Perceptual coding
suppresses sounds that are masked by other sounds. George either does
not understand how perceptual coding works or chooses to obfuscate
what is going on.

[Audio Empire]

On Fri, 15 Oct 2010 11:38:20 -0700, H Davis wrote
(in article ):

"Audio Empire" wrote in message=20
...
=20
snip
=20
Massenburg then used a demonstration to drive his point home. He
electronically subtracted the MP3 compressed music from the original
24-bit/96 KHz recording and then played ONLY the difference signal whi=
ch=20
was
comprised solely of the information lost by the compression.
=20
"These are distortion levels of 15 =EF=BF=BD 20 percent! ", he said as=
he played=20
the
difference signal for all to hear. The distortion amazed everyone in
attendance because it was a grotesquely, but very recognizable version=
of=20
the
original recoding!
=20
He went on to say that while AAC was clearly better than MP3, it still
generated 5% to 10% distortion.
=20
snip
=20
I have just one question. If indeed distortion levels of the magnitudes=
=20
quoted above are present, WHY ARE THEY NOT AUDIBLE? One can certainly h=
ear=20
even one or two percent harmonic distortion.=20
=20
=20

They are audible.=20

[Audio Empire]

On Fri, 15 Oct 2010 15:31:34 -0700, jwvm wrote
(in article ):

On Oct 15, 2:38=A0pm, "H Davis" wrote:
"Audio Empire" wrote in message

...

snip

Massenburg then used a demonstration to drive his point home. He
electronically subtracted the MP3 compressed music from the original
24-bit/96 KHz recording and then played ONLY the difference signal whic=
h
was
comprised solely of the information lost by the compression.

"These are distortion levels of 15 20 percent! ", he said as he played
the
difference signal for all to hear. The distortion amazed everyone in
attendance because it was a grotesquely, but very recognizable version =
of
the
original recoding!

He went on to say that while AAC was clearly better than MP3, it still
generated 5% to 10% distortion.

snip

I have just one question. If indeed distortion levels of the magnitudes
quoted above are present, WHY ARE THEY NOT AUDIBLE? One can certainly hea=
r
even one or two percent harmonic distortion.

Because the distortion is primarily linear. Perceptual coding
suppresses sounds that are masked by other sounds. George either does
not understand how perceptual coding works or chooses to obfuscate
what is going on.

Mmmmm. Are you sure that perceptual coding produces linear distortion? I
thought linear distortion was a filter function where gain (or delay) varies
with frequency.

[Arny Krueger]

"Audio Empire" wrote in message


Mmmmm. Are you sure that perceptual coding produces
linear distortion?

Yes.

I thought linear distortion was a
filter function where gain (or delay) varies with
frequency.

Perceptual coders are lagely based on filter banks.

[Audio Empire]

On Sun, 17 Oct 2010 09:11:33 -0700, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in message


Mmmmm. Are you sure that perceptual coding produces
linear distortion?

Yes.

I thought linear distortion was a
filter function where gain (or delay) varies with
frequency.

Perceptual coders are lagely based on filter banks.



MMMM, I didn't realize that they worked that way. Makes sense, I guess.

[Sebastian Kaliszewski]

Audio Empire wrote:
[...] These people easily hear the difference between CD quality
and high-resolution audio. A DBT test made of such people, people who's ears
I know and respect, who told me that it was statistically impossible for THEM
them to hear the difference between the various sample rates and bit depths,
that would carry no small amount of weight with me. Also, like I said before,
if I were ever privy to a correctly set-up and executed DBT of this type, and
couldn't, in any statistically meaningful way, tell my own 24/192 recordings
from 16/44.1 copies of those recordings, then I would be convinced, and you'd
never hear me even mention high-resolution digital recording again. [...]

Well, the software to perform such test is AFAIR available for free. One
needs 24/192 recording, reduce it down to 16/44.1, upsample back to
24/192[*] (the information lost will to be restored) and comprare
original with processed audio using some ABX software.

No need for some expensive vinyl equipment, etc... Just some good 24/192
recording, and as I understand you have such recordings handy.

*] - upsampling step requires explanation -- some audio equipment can't
promptly and hands free switch sample rates -- that way there will be no
such problem, playbacksettings will be the same, etc. It isolates pure
16/44.1 vs 24/192 difference

rgds
\SK
--
"Never underestimate the power of human stupidity" -- L. Lang
--
http://www.tajga.org -- (some photos from my travels)

[Harry Lavo]

"Arny Krueger" wrote in message
...
"Audio Empire" wrote in message

On Thu, 14 Oct 2010 06:41:37 -0700, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in
message


snip

Every audio enthusiast I know has a decent vinyl playback
system.

I seriously doubt that the quality of all of them would meet the standards
of the most agressive advocates of so-called "Hi rez" audio.

Most the audio older audio enthusiasts I know have vinyl playback systems
too, but most of them are in storage.

The younger enthusiasts I know don't even have decent component stereo
systems.


A few of the younger ones have only recently
bought them too. Even the New York Times has noticed that
there is vinyl renaissance going-on. Haven't you?

Believe it or not PR releases masquerading as NYT news items don't have
much credibility in the heartland.

The only current vinyl renaissance that I know of being represented by
actual hardware involves sub-$200 playback systems which covers most of
the
new LP player hardware purchases that I am aware of.


Well, I guess my nephew wasn't counted then. He is a producer at MTV, and
currently has a Thorens 318 and is soon to acquire a SOTA Star Saphire (all
used, of course). He has not bought a CD for two years...only vinyl, and
most of it is new. His reason: "it just sounds much better".

[glenbadd]

On Oct 19, 1:11=A0am, Sebastian Kaliszewski
wrote:
Well, the software to perform such test is AFAIR available for free. One
needs 24/192 recording, reduce it down to 16/44.1, upsample back to
24/192[*] (the information lost will to be restored) and comprare
original with processed audio using some ABX software.

No need for some expensive vinyl equipment, etc... Just some good 24/192
recording, and as I understand you have such recordings handy.

*] - upsampling step requires explanation -- some audio equipment can't
promptly and hands free switch sample rates -- that way there will be no
such problem, playbacksettings will be the same, etc. It isolates pure
16/44.1 vs 24/192 difference

Upsampling and downsampling generally uses non-linear
algorithms, such as sinc and polyphase, so your comparison
is almost guaranteed to be different.
Ref. http://leute.server.de/wilde/resample.html

G.

[Sebastian Kaliszewski]

glenbadd wrote:
On Oct 19, 1:11 am, Sebastian Kaliszewski
wrote:
Well, the software to perform such test is AFAIR available for free. One
needs 24/192 recording, reduce it down to 16/44.1, upsample back to
24/192[*] (the information lost will to be restored) and comprare
original with processed audio using some ABX software.

No need for some expensive vinyl equipment, etc... Just some good 24/192
recording, and as I understand you have such recordings handy.

*] - upsampling step requires explanation -- some audio equipment can't
promptly and hands free switch sample rates -- that way there will be no
such problem, playbacksettings will be the same, etc. It isolates pure
16/44.1 vs 24/192 difference

Upsampling and downsampling generally uses non-linear
algorithms, such as sinc and polyphase,

So what? Sampling (AD conversion) is non linear by itself. The whole
point is to keep non-linearities in the restored analog signal down to
some low dB. Proper ADC - DAC chain will hold those very close to
theoretical minimum possible.

And it is the same with resampling. Proper resampling (and virtually all
quality audio editting software does it the propoer way) keeps
nonlinearities where they should be (i.e. down at theoretical minimums).

? so your comparison
is almost guaranteed to be different.

Nope.

This is the very nature of difference between 16/44.1 and 24/192 (or any
two different sets of bit depth-sample rate. In fact this is the way
oversampling DACs work -- signal is upsampled and then DA converted and
then simpler restoration filters could be used.

Ref. http://leute.server.de/wilde/resample.html

I'm affraid You're confusing linear interpolation of sample points and
linear transformation (distortion, filter) in signal processing. Linear
interpolation of sample points is a non linear transformation of signal.
It's like confusing piano, an instrument, with piano -- quiet passage in
music -- one can play forte on a piano

rgds
\SK
--
"Never underestimate the power of human stupidity" -- L. Lang
--
http://www.tajga.org -- (some photos from my travels)

[Andrew Haley]

glenbadd wrote:
On Oct 19, 1:11?am, Sebastian Kaliszewski
wrote:
Well, the software to perform such test is AFAIR available for free. One
needs 24/192 recording, reduce it down to 16/44.1, upsample back to
24/192[*] (the information lost will to be restored) and comprare
original with processed audio using some ABX software.

No need for some expensive vinyl equipment, etc... Just some good 24/192
recording, and as I understand you have such recordings handy.

*] - upsampling step requires explanation -- some audio equipment can't
promptly and hands free switch sample rates -- that way there will be no
such problem, playbacksettings will be the same, etc. It isolates pure
16/44.1 vs 24/192 difference

Upsampling and downsampling generally uses non-linear algorithms,
such as sinc and polyphase, so your comparison is almost guaranteed
to be different.

No, they generally use *linear* algorithms such as sinc. (To be more
precise, a brickwall linear phase FIR filter.) There's no reason to
use anything nonlinear unless you want something extremely quick 'n
dirty, and I presume that for this test we'd want the resampling to be
exemplary.

Andrew.

[Arny Krueger]

"glenbadd" wrote in message

On Oct 19, 1:11 am, Sebastian Kaliszewski
wrote:
Well, the software to perform such test is AFAIR
available for free. One needs 24/192 recording, reduce
it down to 16/44.1, upsample back to 24/192[*] (the
information lost will to be restored) and comprare
original with processed audio using some ABX software.

No need for some expensive vinyl equipment, etc... Just
some good 24/192 recording, and as I understand you have
such recordings handy.

*] - upsampling step requires explanation -- some audio
equipment can't promptly and hands free switch sample
rates -- that way there will be no such problem,
playbacksettings will be the same, etc. It isolates pure
16/44.1 vs 24/192 difference

Upsampling and downsampling generally uses non-linear
algorithms, such as sinc and polyphase, so your comparison
is almost guaranteed to be different.
Ref. http://leute.server.de/wilde/resample.html

Do you think that these differences are such that they will invalidate the
comparison?

IOW, are they guaranteed to always produce audible differences?

[glenbadd]

On Oct 19, 1:11=A0am, Sebastian Kaliszewski
wrote:

Well, the software to perform such test is AFAIR available for free. One
needs 24/192 recording, reduce it down to 16/44.1, upsample back to
24/192[*] (the information lost will to be restored) and comprare
original with processed audio using some ABX software.

OK, I'm able to try that. Should I expect all the content above
22050Hz
to be missing from the upsampled 24/192? And below 22050 Hz the
waveforms and spectrums should appear exactly the same ?
(compared to the 24/192 original)
What happens to the original 24 bit information that is below the
16 bit dynamic range of 16/44.1 ?

G.

[Sebastian Kaliszewski]

glenbadd wrote:
On Oct 19, 1:11=A0am, Sebastian Kaliszewski
wrote:
Well, the software to perform such test is AFAIR available for free. One
needs 24/192 recording, reduce it down to 16/44.1, upsample back to
24/192[*] (the information lost will to be restored) and comprare
original with processed audio using some ABX software.

OK, I'm able to try that. Should I expect all the content above
22050Hz
to be missing from the upsampled 24/192?

Yes, after upsampling from 16/44.1 back to 24/192 no original high
frequency information would be there. With good[*] upsampling algorithm
there would be very little newly introduced information (i.e. noise).
With bad algorithm there could be a lot of (introduced) noise up there.

And below 22050 Hz the
waveforms and spectrums should appear exactly the same ?
(compared to the 24/192 original)

Very close, but not exactly the same, as filtering at the cutoff is
performed and it's not 100% brickwall[**]. And with the difference that
low level (below 16bit) information is lost.

Don't worry -- most todays DAC upsample signal -- the do becaus that's a
bettre way to do it as analog reconstruction filteriong (on the analog
signal side of the DAC) is much simpler when band for filter slope is 4
times wider than passband instead if being just 1/10 width of the passband.

What happens to the original 24 bit information that is below the
16 bit dynamic range of 16/44.1 ?

Generally speaking it is lost. What one could do is to use noise shaping
while reducing bit depth and then in the most important band (0.5-6KHz)
dynamic range could be improved somewhat (about 10dB) at the cost of
reduced dynamic range at the extremes of the band.


The whole point of the excercise is to determine if that information
loss is audible. While you're doing that you could also try experiments
with intermediate caseses, i.e. 24/44.1 and 16/192 (to distinguish bit
depth effects vs sample rate effects)

rgds
\SK

[*] - It's possible to devise upsampling algorithm which adds specially
crafted data above the cutoff frequency -- similar to some photoediting
software does while zooming. But this data is still a distortion it's
only designed to please people perceiving it (i.e. "eugraphic" in case
of photos or euphonic i case of audio distortion). It's not the original
data, it's just a fake, a guesstimate how that original data could
more-or-less (rather less than more) look. Original data *is* still *lost*.
[**] - It's theoretically posible to perform excat 100% brickwall in
digital domain but it's costly and that cost highly depends on exact
proportion of starting and ensing frequency. In case of 24/192 to
16/44.1 reduction memory need is 147 times (+ some for calculation data)
the memory needed for original 24/192 data.
For 10s stereo piece at 24/192 there are about 100billion operations to
perform and memory requirement is about 2.2GB.
For 15min track the requirements are two order of magnitude higher
(about 10 trillion operations on about 200GB memory set -- i.e.
unfeasible on todays computers available for general use).
Yet in case of 24/192 to 16/48 it's much better as memory need is just 4
times not 147 times more (as I said -- it strongly depends on exact
proportion of sampling frequencies involved).

--
"Never underestimate the power of human stupidity" -- L. Lang
--
http://www.tajga.org -- (some photos from my travels)

[Andrew Haley]

glenbadd wrote:
On Oct 19, 1:11=A0am, Sebastian Kaliszewski
wrote:

Well, the software to perform such test is AFAIR available for free. One
needs 24/192 recording, reduce it down to 16/44.1, upsample back to
24/192[*] (the information lost will to be restored) and comprare
original with processed audio using some ABX software.

OK, I'm able to try that. Should I expect all the content above
22050Hz to be missing from the upsampled 24/192?

Yes.

And below 22050 Hz the waveforms and spectrums should appear exactly
the same ?

Yes.

(compared to the 24/192 original)
What happens to the original 24 bit information that is below the 16
bit dynamic range of 16/44.1 ?

It's still there, but buried in a raised noise floor. [1]

Andrew.

[1] Digital Dither: Processing with Resolution Far Below the Least
Significant Bit, Vanderkooy, John; Lip****z, Stanley P., 7th
International Conference: Audio in Digital Times (May 1989)

[Arny Krueger]

"glenbadd" wrote in message


On Oct 19, 1:11=A0am, Sebastian Kaliszewski
wrote:

Well, the software to perform such test is AFAIR
available for free. One needs 24/192 recording, reduce
it down to 16/44.1, upsample back to 24/192[*] (the
information lost will to be restored) and comprare
original with processed audio using some ABX software.

OK, I'm able to try that. Should I expect all the
content above 22050Hz to be missing from the upsampled 24/192?

Yes. There should be a near-total elimination of musical information above
22.05 KHz after it is downsampled to 16/44.1

If you want to evaluate the performance of the resampler that you are using,
you might find more information about it he

http://src.infinitewave.ca/help.html

http://www.mainly.me.uk/resampling/index.html



And below 22050 Hz the waveforms and spectrums should appear exactly the
same ? (compared to the 24/192 original)

The results of eliminating all signals above 22.05 KHz on a technical
analysis (e.g. FFT analysis, visual waveform inspection) can range from
trivial to dramatic, depending on the 24/192 recording that you choose.

However, the subjective consequences of eliminating all signals above 22.05
KHz are moot because of how music and overtones in the 12-20 KHz range masks
music and overtones above 20 KHz. Masking essentially deafens you to
sounds that you might hear were they presented in the absence of the masking
sounds.

Masking is the primary reason why MP3 coding works at all. If there were no
masking by the human ear, MP3 encoded music might sound so unlike music that
you wouldn't be able to recognize a familiar performance. While some people
are critical of even high bitrate MP3 encoded music, I've never heard
anybody say that it trashed the music so badly that they couldn't recognize
a familiar performance.

What happens to the original 24 bit information that is
below the 16 bit dynamic range of 16/44.1 ?

As a rule, there simply isn't any 24 bit information below the dynamic range
of 16 bit recording. This is because noise that was picked up (primarily
environmental) during the origional recording process has rased the 24/192
recording's noise floor well above the approximate -96 dB noise floor of
16/44 digital.

The widest commerical dynamic range of *any* commercial recording I have
been able to obtain is about 86 dB. Most are in the low 70's, or even
worse. Thus the noise floor of almost all real-world music is so high that
it competely masks the noise floor that is inherent in 16 bit recording.

[Audio Empire]

On Fri, 22 Oct 2010 07:49:51 -0700, Sebastian Kaliszewski wrote
(in article ):

glenbadd wrote:
On Oct 19, 1:11=A0am, Sebastian Kaliszewski
wrote:
Well, the software to perform such test is AFAIR available for free. One
needs 24/192 recording, reduce it down to 16/44.1, upsample back to
24/192[*] (the information lost will to be restored) and comprare
original with processed audio using some ABX software.

OK, I'm able to try that. Should I expect all the content above
22050Hz
to be missing from the upsampled 24/192?

Yes, after upsampling from 16/44.1 back to 24/192 no original high
frequency information would be there. With good[*] upsampling algorithm
there would be very little newly introduced information (i.e. noise).
With bad algorithm there could be a lot of (introduced) noise up there.

And below 22050 Hz the
waveforms and spectrums should appear exactly the same ?
(compared to the 24/192 original)

Very close, but not exactly the same, as filtering at the cutoff is
performed and it's not 100% brickwall[**]. And with the difference that
low level (below 16bit) information is lost.

Don't worry -- most todays DAC upsample signal -- the do becaus that's a
bettre way to do it as analog reconstruction filteriong (on the analog
signal side of the DAC) is much simpler when band for filter slope is 4
times wider than passband instead if being just 1/10 width of the passband.

What happens to the original 24 bit information that is below the
16 bit dynamic range of 16/44.1 ?

Generally speaking it is lost. What one could do is to use noise shaping
while reducing bit depth and then in the most important band (0.5-6KHz)
dynamic range could be improved somewhat (about 10dB) at the cost of
reduced dynamic range at the extremes of the band.


The whole point of the excercise is to determine if that information
loss is audible. While you're doing that you could also try experiments
with intermediate caseses, i.e. 24/44.1 and 16/192 (to distinguish bit
depth effects vs sample rate effects)

rgds
\SK


[*] - It's possible to devise upsampling algorithm which adds specially
crafted data above the cutoff frequency -- similar to some photoediting
software does while zooming. But this data is still a distortion it's
only designed to please people perceiving it (i.e. "eugraphic" in case
of photos or euphonic i case of audio distortion). It's not the original
data, it's just a fake, a guesstimate how that original data could
more-or-less (rather less than more) look. Original data *is* still *lost*.
[**] - It's theoretically posible to perform excat 100% brickwall in
digital domain but it's costly and that cost highly depends on exact
proportion of starting and ensing frequency. In case of 24/192 to
16/44.1 reduction memory need is 147 times (+ some for calculation data)
the memory needed for original 24/192 data.
For 10s stereo piece at 24/192 there are about 100billion operations to
perform and memory requirement is about 2.2GB.
For 15min track the requirements are two order of magnitude higher
(about 10 trillion operations on about 200GB memory set -- i.e.
unfeasible on todays computers available for general use).
Yet in case of 24/192 to 16/48 it's much better as memory need is just 4
times not 147 times more (as I said -- it strongly depends on exact
proportion of sampling frequencies involved).



My experience is that the info above 22050 Hz is largely irrelevant. The
things that 24/96 and 24/192 bring to the party are better image specificity,
smoother high-frequency reproduction (5K -up to the limits of
audibility-whatever they might be for the individual listener), better
low-level and ambience detail. If all the DBT tests associated with so-called
hi-res audio are concerned with, or largely confined to the effects of the
ultrasonic performance on the actual tonal balance, or other
frequency-response related issues, then those constructing these tests are,
IMHO, barking up the wrong tree by listening for the wrong things in their
tests.

What I have not done is to note any difference between soundstage, imaging,
and low-level information retrieval between 24/96 and 24/192, much less
24/88.2 or even 24/48 or 24/44.1. That's because I've ONLY tested 16/44.1 vs
24/192 for these things. Fort all I know, the higher sampling rate MIGHT be
totally irrelevant (then again, it may not be).

[Arny Krueger]

"Audio Empire" wrote in message


My experience is that the info above 22050 Hz is largely
irrelevant.

The comments below completely invalidate the above.

The things that 24/96 and 24/192 bring to the
party are better image specificity, smoother
high-frequency reproduction (5K -up to the limits of
audibility-whatever they might be for the individual
listener), better low-level and ambience detail.

There's no reason to believe that high sampling rates than 44.1 KHz have
*any* effect, either measured or heard, on imaging, ambience, or low level
detail.

If all the DBT tests associated with so-called hi-res audio are
concerned with, or largely confined to the effects of the
ultrasonic performance on the actual tonal balance, or
other frequency-response related issues, then those
constructing these tests are, IMHO, barking up the wrong
tree by listening for the wrong things in their tests.

Given the vast number of DBTs comparing higher sample rates to 16/44 that
have been done with the null outcomes, there simply is no contrary opinion
that is humble. Besides, there is no reason to weigh this issue based on
opinion when reliable, factual information is so easy to obtain.

[Audio Empire]

On Sat, 23 Oct 2010 13:29:23 -0700, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in message


My experience is that the info above 22050 Hz is largely
irrelevant.

The comments below completely invalidate the above.


Make up your mind, Mr. Kruger. You cannot have it both ways. Either higher
sampling rates (and the concurrent extension of frequency response above 22
KHz that accompanies them) is relevant or it isn't. I say that it's probably
not the extension into the ultra-sonic range that is important here and you
contradict me saying that my above comment is "invalidated"..

The things that 24/96 and 24/192 bring to the
party are better image specificity, smoother
high-frequency reproduction (5K -up to the limits of
audibility-whatever they might be for the individual
listener), better low-level and ambience detail.

There's no reason to believe that high sampling rates than 44.1 KHz have
*any* effect, either measured or heard, on imaging, ambience, or low level
detail.

Then you turn right around and say that sampling rates higher than 44.1 KHz
(with it's concomitant 22.05Khz upper frequency response limit) has no
effect.

Now it either does have an effect or it doesn't. Above you tell me that I'm
wrong and that my first comment is invalidated by "Comments Below" indicating
that you disagree with me when I say that frequency response above 22.05 KHz
is probably irrelevant. Then, in your very next comment, you assert the exact
same thing that I was asserting????? Sometimes I think that you argue here
just to be contrary.


BTW, I never said or implied that the sampling rate was responsible for the
improved imaging, ambience and low-level retrieval. I merely stated that all
of my experiments on this phenomenon were carried out at the higher sampling
rates. In fact, in the very post that you are (selectively) quoting here, I
clearly said that although all my experiments along these lines were done at
either 96 or 192KHz sampling rates, that it was very possible that it IS NOT
the high sampling rate that is responsible for these audible improvements,
but that one might find that it's the 24-bit word length that is responsible
and that 24/48 or 24/44.1 might just yield identical results to 24/96 and
24/192 in this respect.

If all the DBT tests associated with so-called hi-res audio are
concerned with, or largely confined to the effects of the
ultrasonic performance on the actual tonal balance, or
other frequency-response related issues, then those
constructing these tests are, IMHO, barking up the wrong
tree by listening for the wrong things in their tests.

Given the vast number of DBTs comparing higher sample rates to 16/44 that
have been done with the null outcomes, there simply is no contrary opinion
that is humble. Besides, there is no reason to weigh this issue based on
opinion when reliable, factual information is so easy to obtain.

Since I have never insisted that the sample-rate is responsible for
improvements in imaging, ambience, or low level detail and said so in the
very post of mine that you are quoting out of context, your above comment
seems more than a little self serving, misleading and some might say,
mean-spirited.

[Arny Krueger]

"Audio Empire" wrote in message

On Sat, 23 Oct 2010 13:29:23 -0700, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in
message

My experience is that the info above 22050 Hz is largely
irrelevant.

The comments below completely invalidate the above.

Make up your mind, Mr. Kruger. You cannot have it both
ways. Either higher sampling rates (and the concurrent
extension of frequency response above 22 KHz that
accompanies them) is relevant or it isn't.

This is just rhetoric. I've never ever said that information above 22 KHz is
relevant to listening to music. There are reasons to occasionally use higher
sample rates for technical reasons, but no reasons to bother distributing
media with sounds on it that contribute zero to the actual listening
experience.

I say that
it's probably not the extension into the ultra-sonic
range that is important here and you contradict me saying
that my above comment is "invalidated"..

Its not the dynamic range, either.

The things that 24/96 and 24/192 bring to the
party are better image specificity, smoother
high-frequency reproduction (5K -up to the limits of
audibility-whatever they might be for the individual
listener), better low-level and ambience detail.

There's no reason to believe that high sampling rates
than 44.1 KHz have *any* effect, either measured or
heard, on imaging, ambience, or low level detail.

Then you turn right around and say that sampling rates
higher than 44.1 KHz (with it's concomitant 22.05Khz
upper frequency response limit) has no effect.


The two statements exactly agree. Where't the beef?

Now it either does have an effect or it doesn't.

Both statements in my reading say it doesn't.

Above
you tell me that I'm wrong and that my first comment is
invalidated by "Comments Below" indicating that you
disagree with me when I say that frequency response above
22.05 KHz is probably irrelevant. Then, in your very next
comment, you assert the exact same thing that I was
asserting????? Sometimes I think that you argue here
just to be contrary.

I don't know how you can read this from what I wrote.

BTW, I never said or implied that the sampling rate was
responsible for the improved imaging, ambience and
low-level retrieval. I merely stated that all of my
experiments on this phenomenon were carried out at the
higher sampling rates. In fact, in the very post that you
are (selectively) quoting here, I clearly said that
although all my experiments along these lines were done
at either 96 or 192KHz sampling rates, that it was very
possible that it IS NOT the high sampling rate that is
responsible for these audible improvements, but that one
might find that it's the 24-bit word length that is
responsible and that 24/48 or 24/44.1 might just yield
identical results to 24/96 and 24/192 in this respect.


I seriously doubt that any actual improved imaging was ever found. I see no
evidence that was gathered with the kind of care that should be used when
trying to study this kind of far-reaching question.

[Audio Empire]

On Mon, 25 Oct 2010 07:10:17 -0700, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in message

On Sat, 23 Oct 2010 13:29:23 -0700, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in
message

My experience is that the info above 22050 Hz is largely
irrelevant.

The comments below completely invalidate the above.

Make up your mind, Mr. Kruger. You cannot have it both
ways. Either higher sampling rates (and the concurrent
extension of frequency response above 22 KHz that
accompanies them) is relevant or it isn't.

This is just rhetoric. I've never ever said that information above 22 KHz is
relevant to listening to music. There are reasons to occasionally use higher
sample rates for technical reasons, but no reasons to bother distributing
media with sounds on it that contribute zero to the actual listening
experience.


When I said that info above 22050 Hz is largely irrelevant, you said that I
was wrong. How is that rhetoric?

I say that
it's probably not the extension into the ultra-sonic
range that is important here and you contradict me saying
that my above comment is "invalidated"..

Its not the dynamic range, either.

I say it has *something* to do with the higher bit-depth. You say not. WE'll
have to agree to disagree.

The things that 24/96 and 24/192 bring to the
party are better image specificity, smoother
high-frequency reproduction (5K -up to the limits of
audibility-whatever they might be for the individual
listener), better low-level and ambience detail.

There's no reason to believe that high sampling rates
than 44.1 KHz have *any* effect, either measured or
heard, on imaging, ambience, or low level detail.

Then you turn right around and say that sampling rates
higher than 44.1 KHz (with it's concomitant 22.05Khz
upper frequency response limit) has no effect.


The two statements exactly agree. Where't the beef?

Because when I stated that I find the response above 22050 Hz to be largely
irrelevant, you came back and said that (and I quote) "The comments below
completely invalidate the above." Sounds to me like you are disagreeing with
me in that statement and then you say just the opposite several sentences on.
That's the beef.


Now it either does have an effect or it doesn't.

Both statements in my reading say it doesn't.

Then why did you disagree when I said the same thing?

Above
you tell me that I'm wrong and that my first comment is
invalidated by "Comments Below" indicating that you
disagree with me when I say that frequency response above
22.05 KHz is probably irrelevant. Then, in your very next
comment, you assert the exact same thing that I was
asserting????? Sometimes I think that you argue here
just to be contrary.

I don't know how you can read this from what I wrote.

Easy. And again I quote:

I said: "My experience is that the info above 22050 Hz is largely
irrelevant."

Then you responded: "The comments below completely invalidate the above."

If that doesn't constitute a direct refute of my comment, then one of us is
using another language other than English here, and it isn't me.


BTW, I never said or implied that the sampling rate was
responsible for the improved imaging, ambience and
low-level retrieval. I merely stated that all of my
experiments on this phenomenon were carried out at the
higher sampling rates. In fact, in the very post that you
are (selectively) quoting here, I clearly said that
although all my experiments along these lines were done
at either 96 or 192KHz sampling rates, that it was very
possible that it IS NOT the high sampling rate that is
responsible for these audible improvements, but that one
might find that it's the 24-bit word length that is
responsible and that 24/48 or 24/44.1 might just yield
identical results to 24/96 and 24/192 in this respect.


I seriously doubt that any actual improved imaging was ever found.

I can't help it if you have so much invested in 16-bit/44.1 being perfect,
that you cannot or will not entertain any other point of view.

I see no
evidence that was gathered with the kind of care that should be used when
trying to study this kind of far-reaching question.

That's your problem, not mine. I'm not alone in this. Many well-known
professional classical recording engineers of my acquaintance agree with me
on this. We didn't invent these phenomena out of thin air. The experiments
that I have performed on this phenomenon are pretty valid. It's easy to hear,
not subtle at all. Copy a good imaging LP or analog master tape to CD, the
imaging becomes vague and the soundstage shrinks in virtual size. Copy the
SAME record or master tape to 24 -bit (and again, I use either 96 KHz or 192
KHZ for this) and the imaging firms up (you can close your eyes and point to
the various instruments and can even tell whether instruments are in front or
in back of the ensemble), is indistinguishable from the source and the
soundstage widens appreciably. You can also, as I said earlier, hear more
ambience (if it's there in the first place) and one can hear greater
low-level detail.

Of course, I don't doubt that some people, who don't know what to listen for,
would miss many of these things and if they weren't present on the program
material used for any given DBT, then no one would hear these things because
they wouldn't be on the recordings to begin with. Most recordings, even most
classical, orchestral recordings, aren't real stereo to begin with, you know.

[Audio Empire]

On Mon, 25 Oct 2010 16:31:19 -0700, ScottW wrote
(in article ):

On Oct 25, 10:14=A0am, Audio Empire wrote:
On Mon, 25 Oct 2010 07:10:17 -0700, Arny Krueger wrote
(in article ):



[ snipped for excessive quoting -- dsr ]



=A0I see no
evidence that was gathered with the kind of care that should be used wh=
en
trying to study this kind of far-reaching question.

That's your problem, not mine. I'm not alone in this. Many well-known
professional classical recording engineers of my acquaintance agree with =
me
on this. We didn't invent these phenomena out of thin air. The experiment=
s
that I have performed on this phenomenon are pretty valid. It's easy to h=
ear,
not subtle at all. Copy a good imaging LP or analog master tape to CD, th=
e
imaging becomes vague and the soundstage shrinks in virtual size. Copy th=
e
SAME record or master tape to 24 -bit =A0(and again, I use either 96 KHz =
or 192
KHZ for this) and the imaging firms up (you can close your eyes and point=
to
the various instruments and can even tell whether instruments are in fron=
t or
in back of the ensemble), is indistinguishable from the source and the
soundstage widens appreciably.

I'm guessing that your recorder has anti-aliasing filters on ADC
inputs that are the issue at lower sample rates.



You can also, as I said earlier, hear more
ambience (if it's there in the first place) and one can hear greater
low-level detail. =A0

What happens if you record with higher sampling, digitally filter,
and then convert to 16/44?

Does recording via DSD and converting to 16/44.1 count? I've done that with
the same result.


I'm not sure the scenario you provide is
valid to condemn 16/44 CD as
a playback medium.

Neither am I, but I'm not going to dismiss the possibility out-of-hand
either. Most recordings don't have this kind of information in them anyway,
being multi-track (I'd call most commercial recordings made since the
introduction of mult-miking/multi-track recording "overproduced" at best, and
a travesty at worst.). I realize that pop music doesn't generally lend itself
to real stereo production techniques, so the things that I find "improved" by
24-bit/96 or 192 KHz simply don't exist on those types of recordings. And,
since most recording done today is pop and rock, I'd say that the improvement
over 16/44.1 would be a difference that makes no difference at all.
Therefore, CD resolution is fine for most music.

[Audio Empire]

On Tue, 26 Oct 2010 05:45:35 -0700, ScottW wrote
(in article ):

On Oct 25, 5:59=A0pm, Audio Empire wrote:
On Mon, 25 Oct 2010 16:31:19 -0700, ScottW wrote
(in article ):





On Oct 25, 10:14=3DA0am, Audio Empire wrote:
On Mon, 25 Oct 2010 07:10:17 -0700, Arny Krueger wrote
(in article ):

=A0 =A0[ snipped for excessive quoting -- dsr ]

=3DA0I see no
evidence that was gathered with the kind of care that should be used =
wh=3D
en
trying to study this kind of far-reaching question.

That's your problem, not mine. I'm not alone in this. Many well-known
professional classical recording engineers of my acquaintance agree wi=
th =3D
me
on this. We didn't invent these phenomena out of thin air. The experim=
ent=3D
s
that I have performed on this phenomenon are pretty valid. It's easy t=
o h=3D
ear,
not subtle at all. Copy a good imaging LP or analog master tape to CD,=
th=3D
e
imaging becomes vague and the soundstage shrinks in virtual size. Copy=
th=3D
e
SAME record or master tape to 24 -bit =3DA0(and again, I use either 96=
KHz =3D
or 192
KHZ for this) and the imaging firms up (you can close your eyes and po=
int=3D
=A0to
the various instruments and can even tell whether instruments are in f=
ron=3D
t or
in back of the ensemble), is indistinguishable from the source and the
soundstage widens appreciably.

I'm guessing that your recorder has anti-aliasing filters on ADC
inputs that are the issue at lower sample rates.

You can also, as I said earlier, hear more
ambience (if it's there in the first place) and one can hear greater
low-level detail. =3DA0

=A0What happens if you record with higher sampling, digitally filter,
and then convert to 16/44? =A0

Does recording via DSD and converting to 16/44.1 count? I've done that wi=
th
the same result.

That would depend on the converter I suppose. It certainly adds an
unnecessary variable to the experiment which may or may not have an
impact on the outcome.

Well, the converter is software. It's called "AudioGate". I have used the
Korg MR1000 DSD recorder for all these experiments. The Korg will master at
DSD resolution (5.6448 MHz, 1-bit) or LPCM resolutions of from 16 or 24
Bit/44.1 KHz resolution all the way up to 24-bit/192 Khz Resolution. I have
tried both LPCM and DSD to capture the analog master tapes and LPs and see no
difference between capture via DSD and down-converting to 24/192, 24/96, and
16/44.1 or capturing via LPCM directly.


=A0I'm not sure the scenario you provide is
valid to condemn 16/44 CD as
a playback medium.

Neither am I, but I'm not going to dismiss the possibility out-of-hand
either. Most recordings don't have this kind of information in them anywa=
y,
being multi-track (I'd call most commercial recordings made since the
introduction of mult-miking/multi-track recording "overproduced" at best,=
and
a travesty at worst.). I realize that pop music doesn't generally lend it=
self
to real stereo production techniques, so the things that I find "improved=
" by
24-bit/96 or 192 KHz simply don't exist on those types of recordings. And=
,
since most recording done today is pop and rock, I'd say that the improve=
ment
over 16/44.1 would be a difference that makes no difference at all.
Therefore, CD resolution is fine for most music.

Is there a commercially available recording you feel has the kind of
information you're referring to?

Oh yes. Try the Mercury Living Presence recording of Stravinsky's "Firebird"
Ballet with Antal Dorati and the London Symphony. Mercury LP# SR90226. Forget
the Phillips released CD of this work. It's a pale shadow of the LP and
images very poorly. There are others, of course. Any DGG recorded in the
early 1960's or late 1950's are wonderful real stereo recordings because they
were recorded using M-S miking technique. Don't have the record number off
the top of my head, but the DGG of Stanislaus Richter with Von Karajan and
the Berlin Philharmonic playing Tchiakovsky's Piano Concerto #1 in B Minor is
a great imaging recording. One thing that really sticks out is that you can
tell a real stereo recording from a multi-track mono recording almost
instantly. British Lyrita recordings are real stereo as are some London
(British Decca -again from the late 50's or early 60's) and many of the
1950's RCA Red Seals are gangbusters in the imaging department. The reason
why most of these recordings are from the late 50's and early 60's is because
the mid-sixties marked the beginning of the "dark ages" of classical
recording when multi-track and multi-mike techniques took over.



I find most studio recordings of pop or even jazz provide a far more
refined illusion of image than any but the most initimate live
venues.

From what you have just said, it sounds to me like you've never actually
heard a true-stereo recording, or if you have, you weren't paying attention.
Multi-track and multi-mike recording techniques such as those used for pop
and jazz do not image realistically at all. The tradition in jazz recording,
going back to the 60's is to do small jazz groups as three-channel mono. IOW,
all instruments are grouped in three channels either on the far left, the
middle, and the far right. There is no depth because each instrument is
generally miked separately and probably to it's own track (although some jazz
recordings were captured directly to three-track tape). Then the record is
cut with some instruments in the left channel, some (the soloist or vocal if
there is one) are potted equally to the left and right channel giving a
center mono image, and some are pan-potted wholly into the right channel.
There is no depth because, in most cases, none was captured. I realize that I
am generalizing here but that's because just about every microphone setup one
can imagine and every recording arrangement from strictly two-track all the
way up to 96 tracks has been done at some point by somebody!

So for the characteristics you mention, it's hard to imagine how
recordings could be further image improved over what already
subjectively has more of the characteristics you describe than live
music.

Studio recordings generally have NONE of the characteristics I am describing.
Pan-potting instruments into position can, at best, place instruments along a
straight-line between the far left and the far right, but gives no depth.
Mostly, instruments are grouped together, left, right, and center. Again,
they have no depth because that combination of phase relationships and level
difference cues that human hearing uses to locate objects in space sonically,
are NOT being captured.

Most of the DVD-A discs I have are improved over the CD to some
degree but still don't measure up to what I consider the best on CD.
So if there is a difference attributable to format, it's still pales
in magnitude to the impact of recording engineers/producers etc.

My point is that when the recording engineer takes the trouble to make a real
stereo recording using either X-Y, MS, ORTF or some other legitimate true
stereo mike setup and either records or releases it in 16-bit/44.1 KHz CD,
much of the stereo information simply doesn't seem to make it through the
process. BUT, analog true stereo recordings transferred to high-resolution
DVD-A, SACD, etc. and digital true stereo recordings recorded using DSD or
high-resolution LPCM DO exhibit the imaging and soundstaging that real,
correctly done stereo permits. When done right, real stereo is actually
spooky. You can turn off the lights, sit in the dark, and with only your
ears, "see" the entire orchestra or ensemble arrayed before you. The strings
on the left, the percussion behind the strings, violas and cellos front
center with the woodwinds behind them and the brass behind THEM (and usually
on risers - yes, you can hear that when it's there)! On the right are the low
brass and the cellos and bass viols. You can actually point at each
instrument in space. It really is exciting and it's what Alan Blumlein
envisioned when he invented stereo miking back in the 1930's.