| Author |
Message |
Chris Hornbeck
Guest
|
 Posted:
Tue Sep 27, 2005 4:43 am Post subject:
Re: What is digital audio? The Neddy Report. |
|
|
On Sat, 24 Sep 2005 22:38:02 -0700, "vinylbigot"
<vinylbigot@comcast.net> wrote:
| Quote: | Thus, mathematically speaking, the digital domain is far less exact than the
analog domain.
Of course, the real level of exactness is determnined in the digital domain
by the rate of sampling and in the analog domain by the maximum frequency
content of the sample stored. 24 bit 196KHz sampling in the studio has a
theoretical Nyquist limit equal to approximately the analog content stored
in the best technologies--about 100HKz. Of course the digital domain has
the further problem true undistorted content for non-steady state signals
(which is required for the infinite filters in the Nyquist limit) that
limits the pure mathematical content--undistorted--to about 50 KHz at 24
bits and 20KHz at 16 bit in the studio example above.
|
There are several conceptual errors above. The most obvious
is that the level of "exactness" in analog or digital storage
is limited by the number set used; ain't true 'cause ain't
relevant.
The more subtle, because seemingly intuitive, is that
distortion increases as the Nyquist limit is approached.
This idea arises in the same way that the idea of
"stairsteps" in digital storage arises. And is equally
incorrect.
An idealized A/D/A conversion, meaning including dither,
produces a *perfect* reproduction of the original
anti-alias-filtered original, plus a small random
noise component.
And, incidentally, the oft-expressed and oft-assumed
idea that distortion increases at lower signal levels
in an idealized A/D/A conversion is also incorrect.
Regarding "resolution", the only useful definition in
a monotonic system is the same as signal-to-noise
ratio. True monotonicity is rare and difficult in the
real world, but we're still talking theory here.
Good fortune,
Chris Hornbeck
|
|
| Back to top |
|
 |
Chris Hornbeck
Guest
|
| Posted:
Tue Sep 27, 2005 10:16 am Post subject:
Re: What is digital audio? The Neddy Report. |
|
|
On Sun, 25 Sep 2005 07:05:58 +0000 (UTC), Stewart Pinkerton
<patent3@dircon.co.uk> wrote:
| Quote: | So what is your definition of
the difference between analog and digital?
In principle, there isn't one.
|
I'd be interested in exploring this if you are.
Sounds intriguing, if not provacative,
Where to begin?
Thanks for (finally!) an interesting topic,
Chris Hornbeck |
|
| Back to top |
|
|
vinylbigot
Guest
|
| Posted:
Thu Sep 29, 2005 4:15 am Post subject:
Re: What is digital audio? The Neddy Report. |
|
|
Good points. I agree with your comment on number sets. Neither are
relevant to music. I was just noting that a pure and unrestrained analog
number set (real numbers) is more exact than integer representation as
proved by infinite theory. Of course, the infinitude of the integer set far
exceeds our capability to reproduce so the point is moot.
Also agree with dynamics not affecting distortion--again this is directly
visible in the theory.
However, there is a disagreement about what constitutes distortion related
to frequency in the case where infinite IIR filters are not used (as in
today's audio) and/or where the transient signal does not always extend to
the boundaries of the filter interval--even if the filter is theoretically
perfect. What is verifiably true is that higher sampling at the time of
recording produces better sounding music--I've not found anyone who
disagrees with that in a blind A/B listening test. Sony labs also came to
the same conclusion which is one of the reasons the Super Audio CD
technology was introduced. If if were true that 44.1KHz sampling produced
the complete undistorted orginal signal bandpass limited to 22KHz, why does
SACD sound better?
I'll try to find the MIT article that deals with this and post a link to it.
Also, there are a number of AES papers on this topic, but those are not free
to be posted.
"Chris Hornbeck" <chrishornbeckremovethis@att.net> wrote in message
news:904hj15af7mfda4rh7i9jnnkt9qlrt6u2d@4ax.com...
| Quote: | On Sat, 24 Sep 2005 22:38:02 -0700, "vinylbigot"
vinylbigot@comcast.net> wrote:
Thus, mathematically speaking, the digital domain is far less exact than
the
analog domain.
Of course, the real level of exactness is determnined in the digital
domain
by the rate of sampling and in the analog domain by the maximum frequency
content of the sample stored. 24 bit 196KHz sampling in the studio has a
theoretical Nyquist limit equal to approximately the analog content stored
in the best technologies--about 100HKz. Of course the digital domain has
the further problem true undistorted content for non-steady state signals
(which is required for the infinite filters in the Nyquist limit) that
limits the pure mathematical content--undistorted--to about 50 KHz at 24
bits and 20KHz at 16 bit in the studio example above.
There are several conceptual errors above. The most obvious
is that the level of "exactness" in analog or digital storage
is limited by the number set used; ain't true 'cause ain't
relevant.
The more subtle, because seemingly intuitive, is that
distortion increases as the Nyquist limit is approached.
This idea arises in the same way that the idea of
"stairsteps" in digital storage arises. And is equally
incorrect.
An idealized A/D/A conversion, meaning including dither,
produces a *perfect* reproduction of the original
anti-alias-filtered original, plus a small random
noise component.
And, incidentally, the oft-expressed and oft-assumed
idea that distortion increases at lower signal levels
in an idealized A/D/A conversion is also incorrect.
Regarding "resolution", the only useful definition in
a monotonic system is the same as signal-to-noise
ratio. True monotonicity is rare and difficult in the
real world, but we're still talking theory here.
Good fortune,
Chris Hornbeck |
|
|
| Back to top |
|
|
Chris Hornbeck
Guest
|
| Posted:
Thu Sep 29, 2005 4:43 am Post subject:
Re: What is digital audio? The Neddy Report. |
|
|
On Wed, 28 Sep 2005 16:15:15 -0700, "vinylbigot"
<vinylbigot@comcast.net> wrote:
| Quote: | Good points. I agree with your comment on number sets. Neither are
relevant to music. I was just noting that a pure and unrestrained analog
number set (real numbers) is more exact than integer representation as
proved by infinite theory.
|
WTF is "infinite theory"? The above is incorrect in several
interpretations.
| Quote: | However, there is a disagreement about what constitutes distortion related
to frequency in the case where infinite IIR filters are not used (as in
today's audio) and/or where the transient signal does not always extend to
the boundaries of the filter interval--even if the filter is theoretically
perfect. What is verifiably true is that higher sampling at the time of
recording produces better sounding music--I've not found anyone who
disagrees with that in a blind A/B listening test. Sony labs also came to
the same conclusion which is one of the reasons the Super Audio CD
technology was introduced. If if were true that 44.1KHz sampling produced
the complete undistorted orginal signal bandpass limited to 22KHz, why does
SACD sound better?
|
For a better perspective on this issue, I'd suggest
the various threads that have run in rec.audio.pro .
"Verifiably true" is a big word, and SACD is apparently
an apples and oranges issue. In these early days of
an exploration, it's especially important to maintain
a proper scepticism.
| Quote: | I'll try to find the MIT article that deals with this and post a link to it.
Also, there are a number of AES papers on this topic, but those are not free
to be posted.
|
Thanks! Please don't feel that I'm coming down on you
or have some agenda. Actually, I have more than fifty
feet of vinyl and a Keith Monks record cleaner here at
the shack, still in use. But there are so many seemingly
intuitive but actually incorrect ideas floating around
that I try to patch up the ones that I can.
Good fortune,
Chris Hornbeck
"I had a perspective on the booing, too. You
can kill somebody with kindness." -Bob Dylan |
|
| Back to top |
|
|
Phil Allison
Guest
|
| Posted:
Thu Sep 29, 2005 4:43 am Post subject:
Re: What is digital audio? The Neddy Report. |
|
|
"vinylbigot"
| Quote: |
If if were true that 44.1KHz sampling produced the complete undistorted
orginal signal bandpass limited to 22KHz, why does SACD sound better?
|
** It doesn't .
SACD is an notorious fraud.
............ Phil |
|
| Back to top |
|
|
Stewart Pinkerton
Guest
|
| Posted:
Thu Sep 29, 2005 5:55 am Post subject:
Re: What is digital audio? The Neddy Report. |
|
|
On Wed, 28 Sep 2005 16:15:15 -0700, "vinylbigot"
<vinylbigot@comcast.net> wrote:
| Quote: | Good points. I agree with your comment on number sets. Neither are
relevant to music. I was just noting that a pure and unrestrained analog
number set (real numbers) is more exact than integer representation as
proved by infinite theory. Of course, the infinitude of the integer set far
exceeds our capability to reproduce so the point is moot.
Also agree with dynamics not affecting distortion--again this is directly
visible in the theory.
However, there is a disagreement about what constitutes distortion related
to frequency in the case where infinite IIR filters are not used (as in
today's audio) and/or where the transient signal does not always extend to
the boundaries of the filter interval--even if the filter is theoretically
perfect. What is verifiably true is that higher sampling at the time of
recording produces better sounding music--I've not found anyone who
disagrees with that in a blind A/B listening test.
|
I've not found anyone who has actually done such a comparison. OTOH,
24/96 has pretty much become the de facto studio standard, and can be
replayed on any DVD player, so perhaps the point is moot.
| Quote: | Sony labs also came to
the same conclusion which is one of the reasons the Super Audio CD
technology was introduced.
|
Which was fatally flawed - hence the move by Sony away from DSD to the
hybrid DSD-Wide for recording, which is simply another flavour of the
high-oversampled low-bit technology now widely used in all digital
audio systems. SACD replay is simply a down-conversion of this to a
lower-resolution single-bit version.
| Quote: | If if were true that 44.1KHz sampling produced
the complete undistorted orginal signal bandpass limited to 22KHz, why does
SACD sound better?
|
That has yet to be agreed in a true blind test comparison.
--
Stewart Pinkerton | Music is Art - Audio is Engineering |
|
| Back to top |
|
|
Chris Hornbeck
Guest
|
| Posted:
Fri Sep 30, 2005 4:43 am Post subject:
Re: What is digital audio? The Neddy Report. |
|
|
On Thu, 29 Sep 2005 17:11:07 -0700, "vinylbigot"
<vinylbigot@comcast.net> wrote:
| Quote: | WTF is "infinite theory"? The above is incorrect in several
interpretations.
The part of math theory that deals with infinite numbers--first codified by
George Cantor.
|
OK, so you mean something like: the infinite set of
integer numbers is a subset of the infinite set of
real numbers.
Fair enough, and true, but irrelevant.
Thanks for your thoughts,
Chris Hornbeck |
|
| Back to top |
|
|
vinylbigot
Guest
|
| Posted:
Fri Sep 30, 2005 4:43 am Post subject:
Re: What is digital audio? The Neddy Report. |
|
|
"Chris Hornbeck" <chrishornbeckremovethis@att.net> wrote in message
news:becmj1tru8klnt6erfr3vjv38ebe14n16l@4ax.com...
| Quote: | On Wed, 28 Sep 2005 16:15:15 -0700, "vinylbigot"
vinylbigot@comcast.net> wrote:
Good points. I agree with your comment on number sets. Neither are
relevant to music. I was just noting that a pure and unrestrained analog
number set (real numbers) is more exact than integer representation as
proved by infinite theory.
WTF is "infinite theory"? The above is incorrect in several
interpretations.
|
The part of math theory that deals with infinite numbers--first codified by
George Cantor. |
|
| Back to top |
|
|
|
|
|
|
Profile
Log in to check your private messages
Log in