[Sebastian Kaliszewski]

Audio Empire wrote:
On Wed, 2 Mar 2011 08:03:43 -0800, Sebastian Kaliszewski wrote
(in article ):

Audio Empire wrote:
On Tue, 1 Mar 2011 14:05:25 -0800, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in message

I can't answer that except to say again, that there is no
reason to expect that any two DACs would sound the same.
Seems very unscientific. Appears to totally reject the well-known and
widely
accepted belief that the ear has thresholds for perception of noise and
distortion.
No, my statement does not in any way reject that premise. However, it is
being found that these thresholds are not fixed and can not only vary
greatly
from individual to individual but can be affected in a single individual by
levels of stress, and other psychological variables of human emotional
response.
Yes, but that does not change the facts that there are limits to those
tresholds anyways. In a similar way, some men could run much faster than
others, especially after proper training, nutrition, preparations before
run, etc. But it doesn't change the reality that about 25mph is absolute
top human speed. And creating a device wchich would outrun every athlete,
even the best ones, is pretty easy.

Many examples in the scientific literature where even very dissimilar DACs
and ADCs were compared without positive results.
While that's true, it doesn't, in and of itself, prove anything. Remember
the
scientific axiom: "absence of evidence is not evidence of absence". While
DBT
results are useful (especially when a positive result is returned or when
negative results proves a physical or mathematical prediction), negative
results that do not support an otherwise provable result, are just that;
negative results.
But in this case we have apropriate "physical or mathematical
prediction"... FR is flat within +/-0.2dB (better than some combinations
of cable+speaker), and in most significant 35Hz-10kHz it's +/-0.1dB, THD+N
is down below -110dB, IMD is down below -100dB, phase response is flat,
ringing is down -80dB, preringing is even less, jitter is below 1ns, etc.


That's not what I meant. I meant that in the case of cables, for instance,
the physics tells us that cannot be any difference between cables and
interconnects and that they can have no "sound". The physics says that this
is so, and the maths performed on any cable for which we have specs
(resistance/ft, capacitance/ft, inductance/ft) allows us to calculate the
impedance of that cable at any frequency. From "DC" to at least 20 KHz, we
can see that any speaker cable and any interconnect on the market , in any of
the lengths commonly used in home audio, that these conductors have
absolutely no effect on the signal passing through them. The DBTs confirm
what we already know.

Now show me the same sort of physics and maths that predicts that all A/D,
D/A and amplifier circuits will sound the same irrespective of design,
component quality, or build quality.

To be exact, in the case of cables physics tells us they're same sounding
if the connections are good, thickness is within sensible range, etc.
Interconnect in which there is a cold bond bewteen a cable and a connector
might sound a bit strange IOW two cables sound the same given their
parameters are within range.

Now physics and psychoacustics predict that all A/D, D/A and applifier
circuits sound the same given their parameters are within range.

Getting withing range is significantly easier in case of cables, of
course. But that doesn't preclude both improperly made cable (there are
such) as well as properly made active component (with parameters withing
range).

[...]
Anybody who can't hear the difference between a Benchmark, an
Antelope, and a DCS Scarlatti DAC/Master clock combo,
simply isn't paying attention.
No, they're simply doing good bias-controlled listening tests.
That's an assumption not in evidence - on several fronts. First, you are
assuming that proper bias controlled tests haven't been performed,
I might remember things wrong and I could miss some test descriptions of
yours, but from what I remember you descrived one DBT of three DAC's you
discuss few paragraphs above. But from what I rememeber, some claimed
results had no statisticaly significant backing. Also, as far as I
remember, statistical analysis iself (not its claimed results) and full
set of each test results of all participants (not just one person( were
not given to you and that person conducting the test was an representative
of a party interested in demonstrating that $$$$$ components are sonically
improved against $$ components.

I heard a difference. between the DACs as did some others. No, I did not get
to see the tabulated results, as I said a few months ago. Whether I or anyone
else "got it right" with any statistical certainty is unknown to me. But the
fact that I found it fairly easy to distinguish one DAC from another has made
me mighty skeptical of the "All DACs sound alike" school.

Since then I have lived with both a DCS Scarlatti/DCS Master Clock box as
well as the Antelope Zodiac, and they do different things to the same digital
source on long term listening (I've already established to my satisfaction in
a DBT that they are "different" and now the long-term testing in my system
tells me what those differences are), and it ain't subtle!.

Well, I've maybe written about this before, but I'm not sure. Anways...
Some (not very long) time ago my friend tried to decide which aplifier to
buy (as his pevious one showed aging problems). He took 3 amplifiers home
and wanted to check them. He also wanted to consult with someone other
which would be the right buy, so invited me to listen to them. So the
informal test begun. We compared amp A with amp B. Amp B seemd to sound
nicer -- "sweeter" and more "musical". So we compared B against C -- again
C sounded nicer again (again "sweeter" and more "musical"). So compared C
against A (to close the cycle). And, funili enough, it was A which souned
nicer. One could repeat the test and it was the same -- next applifier
souned nicer in each compared pair We then reversed order to B vs A and
then it was again, the second one of each tested pair souned nicer. So far
with such kind of evaluation. Results were clearly in our brains (and most
probably short and mid term musical memory) not in the equipment.



I'm not ignoring it. Had I wanted to do that, I would have left out that part
of the quote (as some who post here would have, no doubt, surely done). I
agree that the burden of proof is on those of us who are skeptical of Meyer
and Moran's (or any of the other null-result tests of this particular
premise) result. I just don't know how to go about testing that hypothesis in
a scientific manner. IOW, if one is sure that DBTs aren't reliable for this
kind of test because one is 100% sure that DACs sound different and that
those differences make themselves known only over time, how does one prove it
to the satisfaction of all concerned?

Well, DBT could be performed over long periods of time (even days or
weeks). It needs some preparations, but is doable. Setup two devices being
compared, set their gain to measuredly same level (within 0.1dB), and
connect them to the same source and allow them to work concurrently. Only
put blind randomized switch to their outputs. Turn the random switch at
the beginig of the session/day/week/other listening period, when, after
comfortable time of otherwise normal usage of the sysem, you feel you know
which output of which device is passed by a switch (and which is blocked)
note down your answer and then check what was real switch setting and note
down wether your answer was right or wrong. Something like this:
http://www.amazon.co.uk/product-revi...owViewpoints=1
simple device allows to split digital signal coming from one source to
inputs of two devices. Alternatively, if those DACs do not show the level
of the incoming signal or indicate wether incoming signal is present (or
is non zero) one could switch the inputs as well.

There should be about 15 or more such sessions to get statistically
significant results, so the whole test could take a while, but it doesn't
make it impossible to perform.


rgds
\SK

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

[Audio Empire]

On Fri, 4 Mar 2011 06:29:00 -0800, Sebastian Kaliszewski wrote
(in article ):

Audio Empire wrote:
On Wed, 2 Mar 2011 08:03:43 -0800, Sebastian Kaliszewski wrote
(in article ):

Audio Empire wrote:
On Tue, 1 Mar 2011 14:05:25 -0800, Arny Krueger wrote
(in article ):

"Audio Empire" wrote in message

I can't answer that except to say again, that there is no
reason to expect that any two DACs would sound the same.
Seems very unscientific. Appears to totally reject the well-known and
widely
accepted belief that the ear has thresholds for perception of noise and
distortion.
No, my statement does not in any way reject that premise. However, it is
being found that these thresholds are not fixed and can not only vary
greatly
from individual to individual but can be affected in a single individual
by
levels of stress, and other psychological variables of human emotional
response.
Yes, but that does not change the facts that there are limits to those
tresholds anyways. In a similar way, some men could run much faster than
others, especially after proper training, nutrition, preparations before
run, etc. But it doesn't change the reality that about 25mph is absolute
top human speed. And creating a device wchich would outrun every athlete,
even the best ones, is pretty easy.

Many examples in the scientific literature where even very dissimilar
DACs
and ADCs were compared without positive results.
While that's true, it doesn't, in and of itself, prove anything. Remember
the
scientific axiom: "absence of evidence is not evidence of absence". While
DBT
results are useful (especially when a positive result is returned or when
negative results proves a physical or mathematical prediction), negative
results that do not support an otherwise provable result, are just that;
negative results.
But in this case we have apropriate "physical or mathematical
prediction"... FR is flat within +/-0.2dB (better than some combinations
of cable+speaker), and in most significant 35Hz-10kHz it's +/-0.1dB, THD+N
is down below -110dB, IMD is down below -100dB, phase response is flat,
ringing is down -80dB, preringing is even less, jitter is below 1ns, etc.


That's not what I meant. I meant that in the case of cables, for instance,
the physics tells us that cannot be any difference between cables and
interconnects and that they can have no "sound". The physics says that this
is so, and the maths performed on any cable for which we have specs
(resistance/ft, capacitance/ft, inductance/ft) allows us to calculate the
impedance of that cable at any frequency. From "DC" to at least 20 KHz, we
can see that any speaker cable and any interconnect on the market , in any
of
the lengths commonly used in home audio, that these conductors have
absolutely no effect on the signal passing through them. The DBTs confirm
what we already know.

Now show me the same sort of physics and maths that predicts that all A/D,
D/A and amplifier circuits will sound the same irrespective of design,
component quality, or build quality.

To be exact, in the case of cables physics tells us they're same sounding
if the connections are good, thickness is within sensible range, etc.
Interconnect in which there is a cold bond bewteen a cable and a connector
might sound a bit strange IOW two cables sound the same given their
parameters are within range.

No. The physics and math tell us what the performance characteristics of the
wire is. And of course, measurements will tell us the quality of the
connections.

Now physics and psychoacustics predict that all A/D, D/A and amplifier
circuits sound the same given their parameters are within range.

I disagree. Measurements tell us what some of the performance characteristics
of a electronic circuit will be and the physics and math will characterize
that device to a certain point. We use maths to design these devices, they
tell us, for instance, what resistors to use to bias a transistor for the
correct current flow, and to set the feedback for the gain. Maths tell us,
what size capacitor to use to couple the lowest frequency in which we're
interested from stage to stage. Maths also allow us to tailor filters to our
needs and tell us how they will perform in the frequency domain. What the
physics and maths don't predict at the design level (among other things) is
the difference in many performance parameters between components of different
qualities. For instance, I can design an all transistor amplifier and get all
of the component values right, and yet ruin the design sonically, just by
choosing the wrong kind of component. A high gain stage might call for 33,000
Ohm resistor. OK, fine. I'll use a 33,000 Ohm resistor. But if I choose a
carbon composition resistor instead of a metal film, that high gain stage
will be noisy. The maths and physics I used to design that amplifier didn't
predict that, and if I build TWO such amps, one with metal film resistors and
one with carbon comp resistors, they'll sound different and anyone will
instantly tell them apart in a DBT! Same thing with capacitor selection. If
my design called for a a series of coupling capacitors capacitor in the
signal path and I used tantalum capacitors in those spots instead of a some
kind of low DA film capacitor like a polypropylene or a mylar film
capacitor, the amp circuit is going to sound different than it would had I
used the low DA types of capacitors.

This is not as cut and dry as it seems. While the laws of physics will
predict that the two types of resistors will have very different self-noise
characteristics, that's not generally a primary consideration when designing
an amplifier. Sure, the designer probably knows better than to use certain
components, and what the results would be if he did, but the physics behind
the design exercise don't encompass those types of choices. They only predict
such things as frequency response, gain, harmonic and intermodulation
distortion and signal-to-noise ratio based on the parameters of the
components used. However, change the quality of the components and one can
make two identical amplifier sound different, and that's the point.

Getting withing range is significantly easier in case of cables, of
course.

Yeah, for interconnects it just need to be wire. For speakers, it just needs
to be heavy enough wire for power of the amp used.




But that doesn't preclude both improperly made cable (there are
such)

Yeah like cables with suspicious boxes and bulges built into them suggesting
that they contain components other than just wire and a couple of connectors.
But these are no longer conductors, they are passive filters - fixed tone
controls, as it were.


as well as properly made active component (with parameters withing
range).

But that's a tall order. Electronic components sound different from one
another and lots of them in a big design like a high-power amplifier or even
a preamp add together and change the sound. That's why designers have to
measure and listen to their designs after they've designed and prototyped
them. To hear Arny and some of the others here, One would think that all an
audio designer need do, is draw the design out on the back of a napkin and
say, "put it into production!" I mean if all amps sound the same, why spend
the time and money testing and "tweaking" the design?


Anybody who can't hear the difference between a Benchmark, an
Antelope, and a DCS Scarlatti DAC/Master clock combo,
simply isn't paying attention.
No, they're simply doing good bias-controlled listening tests.
That's an assumption not in evidence - on several fronts. First, you are
assuming that proper bias controlled tests haven't been performed,
I might remember things wrong and I could miss some test descriptions of
yours, but from what I remember you descrived one DBT of three DAC's you
discuss few paragraphs above. But from what I rememeber, some claimed
results had no statisticaly significant backing. Also, as far as I
remember, statistical analysis iself (not its claimed results) and full
set of each test results of all participants (not just one person( were
not given to you and that person conducting the test was an representative
of a party interested in demonstrating that $$$$$ components are sonically
improved against $$ components.

I heard a difference. between the DACs as did some others. No, I did not
get
to see the tabulated results, as I said a few months ago. Whether I or
anyone
else "got it right" with any statistical certainty is unknown to me. But
the
fact that I found it fairly easy to distinguish one DAC from another has
made
me mighty skeptical of the "All DACs sound alike" school.

Since then I have lived with both a DCS Scarlatti/DCS Master Clock box as
well as the Antelope Zodiac, and they do different things to the same
digital
source on long term listening (I've already established to my satisfaction
in
a DBT that they are "different" and now the long-term testing in my system
tells me what those differences are), and it ain't subtle!.

Well, I've maybe written about this before, but I'm not sure. Anways...
Some (not very long) time ago my friend tried to decide which aplifier to
buy (as his pevious one showed aging problems). He took 3 amplifiers home
and wanted to check them. He also wanted to consult with someone other
which would be the right buy, so invited me to listen to them. So the
informal test begun. We compared amp A with amp B. Amp B seemd to sound
nicer -- "sweeter" and more "musical". So we compared B against C -- again
C sounded nicer again (again "sweeter" and more "musical"). So compared C
against A (to close the cycle). And, funili enough, it was A which souned
nicer. One could repeat the test and it was the same -- next applifier
souned nicer in each compared pair We then reversed order to B vs A and
then it was again, the second one of each tested pair souned nicer. So far
with such kind of evaluation. Results were clearly in our brains (and most
probably short and mid term musical memory) not in the equipment.

I'm sure it was in your brain. The point is that's no way to evaluate an
amplifier. The way to evaluate which amp you like best is to replace your
current amp with a new contender and listen to it for a few days. Get to know
how it sounds on a variety of program material. If you find that after a few
hours, any differences that you heard between your old and your new amp have
sort of disappeared in your mind, then the new amp is probably fine. If those
differences still bother you after a few days, then you'd best try another.
The reality is that modern solid-state amps sound much closer to one another
than they ever have in history. This is because there are things that have
become standard practice in amp design, things like no or low global
feedback. The use of low-noise resistors, polypropylene and polystyrene
capacitors throughout, oversized power supplies, strong class-A bias for the
first 10-30 Watts of power, MOSFET output devices, etc.. The result is that
most amps sound very good these days. So good that the differences disappear
from most of our consciousnesses after but a few moments of listening.


I'm not ignoring it. Had I wanted to do that, I would have left out that
part
of the quote (as some who post here would have, no doubt, surely done). I
agree that the burden of proof is on those of us who are skeptical of Meyer
and Moran's (or any of the other null-result tests of this particular
premise) result. I just don't know how to go about testing that hypothesis
in
a scientific manner. IOW, if one is sure that DBTs aren't reliable for this
kind of test because one is 100% sure that DACs sound different and that
those differences make themselves known only over time, how does one prove
it
to the satisfaction of all concerned?

Well, DBT could be performed over long periods of time (even days or
weeks). It needs some preparations, but is doable. Setup two devices being
compared, set their gain to measuredly same level (within 0.1dB), and
connect them to the same source and allow them to work concurrently. Only
put blind randomized switch to their outputs. Turn the random switch at
the beginig of the session/day/week/other listening period, when, after
comfortable time of otherwise normal usage of the sysem, you feel you know
which output of which device is passed by a switch (and which is blocked)
note down your answer and then check what was real switch setting and note
down wether your answer was right or wrong. Something like this:

http://www.amazon.co.uk/product-
reviews/B000RWDUYO/ref=dp_top_cm_cr_acr_txt?ie
=UTF8&showViewpoints=1
simple device allows to split digital signal coming from one source to
inputs of two devices. Alternatively, if those DACs do not show the level
of the incoming signal or indicate wether incoming signal is present (or
is non zero) one could switch the inputs as well.

There should be about 15 or more such sessions to get statistically
significant results, so the whole test could take a while, but it doesn't
make it impossible to perform.

But we still don't have proof that DBTs are reliable for audio, we just
assume they are because they work so well for other kinds of bias-controlled
testing.

[bob]

On Mar 4, 2:35=A0pm, Audio Empire wrote:

But we still don't have proof that DBTs are reliable for audio, we just
assume they are because they work so well for other kinds of bias-control=
led
testing.

To the contrary, we have proof that nothing *except* DBTs are reliable
for audio.

bob

[Audio Empire]

On Fri, 4 Mar 2011 19:24:35 -0800, bob wrote
(in article ):

On Mar 4, 2:35=A0pm, Audio Empire wrote:

But we still don't have proof that DBTs are reliable for audio, we just
assume they are because they work so well for other kinds of bias-control=
led
testing.

To the contrary, we have proof that nothing *except* DBTs are reliable
for audio.

bob


Really, how do we prove that? With DBTs?

[Arny Krueger]

"Audio Empire" wrote in message

On Fri, 4 Mar 2011 06:29:00 -0800, Sebastian Kaliszewski
wrote (in article ):

No. The physics and math tell us what the performance
characteristics of the wire is. And of course,
measurements will tell us the quality of the connections.

The same is true for more complex electronics.

Now physics and psychoacustics predict that all A/D, D/A
and amplifier circuits sound the same given their
parameters are within range.

I disagree. Measurements tell us what some of the
performance characteristics of a electronic circuit will
be and the physics and math will characterize that device
to a certain point.

That point is very well refined.

We use maths to design these devices,
they tell us, for instance, what resistors to use to bias
a transistor for the correct current flow, and to set the
feedback for the gain. Maths tell us, what size capacitor
to use to couple the lowest frequency in which we're
interested from stage to stage. Maths also allow us to
tailor filters to our needs and tell us how they will
perform in the frequency domain.

The mathematics all predict nonlinear distortion, etc.

What the physics and
maths don't predict at the design level (among other
things) is the difference in many performance parameters
between components of different qualities.

Not true. The mathematical models for various qualities and types of
components are known and can be plugged into the circuit models.

For instance,
I can design an all transistor amplifier and get all of
the component values right, and yet ruin the design
sonically, just by choosing the wrong kind of component.

Not a problem for the reasons already stated.

A high gain stage might call for 33,000 Ohm resistor. OK,
fine. I'll use a 33,000 Ohm resistor. But if I choose a
carbon composition resistor instead of a metal film, that
high gain stage will be noisy.

You're joking, right? Nobody is using carbon composition resistors these
days.

The maths and physics I
used to design that amplifier didn't predict that, and if
I build TWO such amps, one with metal film resistors and
one with carbon comp resistors, they'll sound different
and anyone will instantly tell them apart in a DBT!

I remember what life was like in the days of carbon composition resistors,
and I also remember what happened on the occasions where I replaced carbon
comp resistors with metal film resistors. In general: Nothing. The problem
was not so much what a good carbon comp resistor did, its what happened when
that resistor went into some of its possible partial failure moded.

Same
thing with capacitor selection. If my design called for a
a series of coupling capacitors capacitor in the signal
path and I used tantalum capacitors in those spots
instead of a some kind of low DA film capacitor like a
polypropylene or a mylar film capacitor, the amp circuit
is going to sound different than it would had I used the
low DA types of capacitors.

Same story. I even had a well-known capacitor dielectric maven whose named
rhymed with bung send me some good and bad capacitors to try in some
projects. The so-called bad capacitors were simply not the part that long
accepted wisdom said should be used in the application. The good capacitors
were film capacitors but in actual use there was no measuable or audible
benefit as compared again to what long accepted wisdom said should be used.

DA is important in sample-and-hold circuits and afew other applications. The
fallacies associated with audio enthusiast misunderstandings of DA have been
explained well by well-known and highly regarded experts such as Robert
Pease of National Semiconductor.

This is not as cut and dry as it seems. While the laws of
physics will predict that the two types of resistors will
have very different self-noise characteristics, that's
not generally a primary consideration when designing an
amplifier.

Absolute and total misrepesentation of generally accepted engineering
standards, even those long before audiophile capacitor parania struck.

Sure, the designer probably knows better than
to use certain components, and what the results would be
if he did, but the physics behind the design exercise
don't encompass those types of choices. They only predict
such things as frequency response, gain, harmonic and
intermodulation distortion and signal-to-noise ratio
based on the parameters of the components used. However,
change the quality of the components and one can make two
identical amplifier sound different, and that's the
point.

There are no known relevant audio circuit design performance parameters
other than linear distortion (phase and frequency response), nonlinear
distortion (harmonic distortion and IM) and noise. People can pretend what
they want, but any other performance parameters only show up in poorly-done
listening tests. IOW, they are false positives.

[Audio Empire]

On Sat, 5 Mar 2011 08:33:54 -0800, Arny Krueger wrote
(in article ):

A high gain stage might call for 33,000 Ohm resistor. OK,
fine. I'll use a 33,000 Ohm resistor. But if I choose a
carbon composition resistor instead of a metal film, that
high gain stage will be noisy.

You're joking, right? Nobody is using carbon composition resistors these
days.

No, I'm not joking. I was using an extreme case to make a point, and that
SHOULD be obvious to even the most casual observer. Of course, nobody uses
carbon comp resistors any more, but if one did use them throughout an amp
design, that amp would sound different from one using metal film resistors.

Same
thing with capacitor selection. If my design called for a
a series of coupling capacitors capacitor in the signal
path and I used tantalum capacitors in those spots
instead of a some kind of low DA film capacitor like a
polypropylene or a mylar film capacitor, the amp circuit
is going to sound different than it would had I used the
low DA types of capacitors.

Same story. I even had a well-known capacitor dielectric maven whose named
rhymed with bung send me some good and bad capacitors to try in some
projects. The so-called bad capacitors were simply not the part that long
accepted wisdom said should be used in the application. The good capacitors
were film capacitors but in actual use there was no measuable or audible
benefit as compared again to what long accepted wisdom said should be used.

IOW, Walt sent you some tantalums (or maybe some aluminum) electrolytics and
some Polypropylenes? Tantalums shouldn't be used in audio circuits for a
number of reasons, and you are right, the wisdom not use them is well known
and long established, and I know that. But again, this is an extreme example
to show that component type and quality can change the quality of an
otherwise decent design.

DA is important in sample-and-hold circuits and afew other applications. The
fallacies associated with audio enthusiast misunderstandings of DA have been
explained well by well-known and highly regarded experts such as Robert
Pease of National Semiconductor.

I know that he disagrees with Mr. Jung et al on this issue, but blind tests
between two Hafler preamp kits, many years ago, one wired per the factory,
and the other wired with "Wondercaps" in place of the factory supplied
capacitors, showed conclusively that the "Wondercap" wired Hafler sounded
much cleaner than the one wired with the factory caps. That and an experience
where I replaced the Mylar film caps with "Wondercaps" in my Magnaplanar
Tympani 3Cs (the ones with the eight panels) showed me conclusively (as far
as I'm concerned) that Jung was correct about capacitor sound.

[Arny Krueger]

"Audio Empire" wrote in message

On Sat, 5 Mar 2011 08:33:54 -0800, Arny Krueger wrote
(in article ):

A high gain stage might call for 33,000 Ohm resistor.
OK, fine. I'll use a 33,000 Ohm resistor. But if I
choose a carbon composition resistor instead of a metal
film, that high gain stage will be noisy.

You're joking, right? Nobody is using carbon composition
resistors these days.

No, I'm not joking. I was using an extreme case to make a
point, and that SHOULD be obvious to even the most casual
observer. Of course, nobody uses carbon comp resistors
any more, but if one did use them throughout an amp
design, that amp would sound different from one using
metal film resistors.

That is a claim that I have personally falsified. In almost every case a
properly operating carbon comp resistor in audio gear causes no audible
difficulties, either on the grounds of noise or nonlinear distortion. I have
actually replaced every carbon comp resistors in a number of audio
components (amps, preamps) and listened to and measured the differences.

Same
thing with capacitor selection. If my design called for
a a series of coupling capacitors capacitor in the
signal path and I used tantalum capacitors in those
spots instead of a some kind of low DA film capacitor
like a polypropylene or a mylar film capacitor, the amp
circuit is going to sound different than it would had I
used the low DA types of capacitors.

Same story. I even had a well-known capacitor dielectric
maven whose named rhymed with bung send me some good and
bad capacitors to try in some projects. The so-called
bad capacitors were simply not the part that long
accepted wisdom said should be used in the application.
The good capacitors were film capacitors but in actual
use there was no measuable or audible benefit as
compared again to what long accepted wisdom said should
be used.

IOW, Walt sent you some tantalums (or maybe some
aluminum) electrolytics and some Polypropylenes?

He sent me Hi-K ceramics and and polypropylenes to be used as coupling
capacitors.

Tantalums shouldn't be used in audio circuits for a
number of reasons,

This is false. Tantalums cause no audible or measurable problems provided
they have appreciable DC voltage across them. IOW they work well in audio
components with single-ended power supplies.

DA is important in sample-and-hold circuits and afew
other applications. The fallacies associated with audio
enthusiast misunderstandings of DA have been explained
well by well-known and highly regarded experts such as
Robert Pease of National Semiconductor.

I know that he disagrees with Mr. Jung et al on this
issue, but blind tests between two Hafler preamp kits,
many years ago, one wired per the factory, and the other
wired with "Wondercaps" in place of the factory supplied
capacitors, showed conclusively that the "Wondercap"
wired Hafler sounded much cleaner than the one wired with
the factory caps.

I know of only listening tests whose results were contrary to that. Probably
just another poorly-done single blind (i.e., defective double blind)
evaluation.

In fact its hard to find measurable differences in that situation, let alone
audible ones.

That and an experience where I replaced
the Mylar film caps with "Wondercaps" in my Magnaplanar
Tympani 3Cs (the ones with the eight panels) showed me
conclusively (as far as I'm concerned) that Jung was
correct about capacitor sound.

The hidden agenda in upgrades like this is that the capacitance of the
capacitors and other traditional parameters such as ESR no doubt changed.
The ESR of parts like these are calculated into a well-done crossover
design. In the case of one well-known anecdote relating to Maggies, the
original parts were not soldered in, but the replacements were.