A question about bits and LPs
S888Wheel
experts have claimed that redbook CD has in effect more "bits" of
information than LP. I haven't been able to figure out how to search
the records of RAHE so I'm putting the question out. Is this true
that the this claim of "more bits" has been made? If so what is this
based on?
Arny Krueger
news:ajj8v...@enews3.newsguy.com
> It has been claimed that on these message boards some of the
> regular experts have claimed that redbook CD has in effect more
> "bits" of information than LP.
This is true. The equivalent number of "bits" of information in an
analog system are given by the equation:
<http://www.tektronix.com/Measurement/cgi-bin/framed.pl?Document=/Mea
surement/App_Notes/DigitalTV/audio.html&FrameSet=television>
SNR = (6.02 * n) + 1.76
There are a number of variations on this equation where number 6.02
is replaced by 6.0. This is an approximation as the true theoretical
number is easily calculated to be 6.02 dB. There are a number of
variations in this equation where the number 1.76 varies by +/- 0.02.
Obviously, neither of these differences are very significant.
Solving for n:
n = (SNR-1.76)/6.02
Note: This equation is for unweighted SNR.
So what is the actual SNR of vinyl?
I've measured unweighted vinyl SNR with a number of different test
records, turntables, cartridges, tonearms, and preamplifiers. In no
case have I observed vinyl playback with greater than 62 dB
unweighted SNR. That gives vinyl resolution approximately equal to 10
bits.
JohnAFR
>I've measured unweighted vinyl SNR with a number of different test
>records, turntables, cartridges, tonearms, and preamplifiers. In no
>case have I observed vinyl playback with greater than 62 dB
>unweighted SNR. That gives vinyl resolution approximately equal to 10
>bits.
That's very interesting. I'm curious how you measured SNR on vinyl.
Can you describe the procedure, test equipment, etc.
How did you determine the maximum signal level?
Thanks,
John Elison
Richard D Pierce
It is based on the sampling theorem as developed by Claude
Shannon in 1948.
In short, it goes something like this. The amount of information
in any analog channel is proportional to the dynamic range of
the medium times the bandwidth of the medium. The dynamic range
is the ratio between the the maximum level encodable and the
lowest level that is unambiguously separable from the noise
In the case of the LP, the bandwidth is seldom, if ever, in
excess of 20 kHz, and often significantly less than that. Also,
in the case of LP, the dynamic range seldom reaches 75 dB, and
is pretty much physically limited to 80 dB at its theoretical
best, extremely rare counterexamples and claims notwithstanding.
The sampling theorem states that if one wants to perfectly
encode a channel whose bandwidth is f, one needs to sample at a
frequency of greater than 2f to perfectly capture ALL the
information contained within the bandwidth of f. And it also
states that to capture a dynamic range of R dB with a binary
representation, one needs a minimum of roughly R/6.02 bits.
Assuming correct implementation, that would indicate that a
quantized, sampled system with a sampling rate of 2*20 kHz or 40
kHz and a bit depth of 75dB/6.02 bits/db or 12.5 bits matches
the bandwidth and range requirements for high-quality LP
playback. Looking at it another way, considering 2-channel
playback, LP has an information rate of approximately 1
megabits/second. (40 kHz sample rate*2 channels*12.5
bits/channel)
Redbook CD has a 44.1 kHz sample rate giving a bandwidth of 20
kHz or more, a 16 bit width giving a dynamic range of
approximate 96 dB, and two channels, resulting in an information
rate for the audio portion of 1.411 MBits/sec.
There will be those that argue, quite incorrectly, that the
sample rate and resolution of LP is "infinite" and thus the bit
rate is "infinite." These arguments are propped up on completely
cinorrect notions like "continuous" means the same as "infinite
resolution." The resolution in the time domain is a measure of
how often unique and unambiguous changes of state can occur, and
is DIRECTLY related to bandwidth. SImilarily, resolution in the
amplitude domain is a measure of how small a change can be
unabiguously encoded in the presence of noise, and is DIRECTLY
related to dynamic range.
Taken to its absurd logic end, "infinite resolution" in both
time and amplitude requires a system that exists for infinite
time and has available infinite energy and bandwidth. No such
system can evebn exist on a hypothetical basis.
None of the proponets of the "analog is infinite" 'theories' has
advanced a single credible argument to support their contention,
nor have they even shown the possibility of a flaw in the logic
of the sampling theorem.
And none of this suggest what system one person or another might
like to listen to. The specific technical question you posed
above regarding the nit rate, or more properly, the amount of
information in LPs. That technical question has a specific
technicla answer, and Claude Shannon, in his paper "A
Mathematical Theory of Communications," showed a definitive
means of calculating that answer.
--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| DPi...@world.std.com |
Cannot Understand Why You Dis AOL
From: s888...@aol.com (S888Wheel)
>Date: 08/16/2002 12:24 PM Eastern Daylight Time
>Message-id: <ajj8v...@enews3.newsguy.com>
Scott, the links to two RAHE threads that I originally posted to
www.stevehoffman.tv were:
<http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&threadm=6a2jo3%241m7%40
merckx.graphics.cornell.edu&rnum=6&prev=/%20groups%3Fq%3D%2522resolution%2
Bof%2Bvinyl%2522%2Bg<br%20/>roup:rec.audio.high-end%26hl%3Den%26lr%3D%26ie
%3DUTF-%208%26scoring%3Dd%26selm%3D6a2jo3%25241m7%2540merckx<br%20/>.graph
ics.cornell.edu%26rnum%3D6>
and
<http://groups.google.com/groups?q=%22resolution+of+vinyl%22+group:rec.audi
o.high-end&hl=en&lr=&ie=UTF-8&scoring=d&selm=51p42b%24reb%40eyrie.graphics
.cornell.edu&rnum=8>
_______
-S.
Arny Krueger
news:ajknsi$lst$1...@bourbaki.localdomain
> "Arny Krueger" <ar...@hotpop.com> wrote:
The best tool I've found for measuring SNR is the relatively new High
Fidelity News (HFN) test record. I establish maximum output level
using its trackability test tracks. I set the maximum level as the
loudest one that plays with a visibly unclipped waveform on a 'scope,
which corresponds to something like 1-3% THD, based on measurements.
The turntable has been adjusted for best possible performance with
these tracks.
I have a fair collection of test records including the latest Shure
Trackability record, various CBS labs records, the Cardas test
record, the Stereo review test record, and of course the HFN test
record. I believe I have at least 2 of each, at least one copy is
kept in reserve for use if I doubt the results from the "working"
copy.
I use a quiet groove to establish the minimum level. BTW, I have a
number of copies of the various test records I use and have reserved
a brand new HFN disc that is used only for this purpose. It will be
reassigned when I think it is too noisy. The preamps used for the
tests have at least 10 dB better measured noise performance with the
stylus up or on a motionless record, compared to what is measured
with the stylus on the moving record.
All measurements are made using a PC with either a Card Deluxe or a
LynxTwo sound card. Either card has > 100 dB unweighted SNR.
Measurement software is Spectra Lab 4.32. The measurement band is
set with - 3 dB points of 20 Hz and 20 KHz with 6 dB/octave rolloff
outside that band. The sample rate is either 96 KHz or 192 KHz so
there is a "brick wall" at either 48 or 96 KHz.
I'd like to make some lacquers of the test tracks I use for my
www.pcabx.com tests. I think most vinylphiles would be blown away by
how easy it is to detect LP and analog tape record/playback in DBTs
as compared to CD record/playback.
TubeGarden
Hmmm. SNR is not the end of the audio trail.
It is possible to hear "into" the noise floor.
And then the sampling frequency isuue, what did you take as the sampling
frequency of vinyl?
Happy Ears!
Al
JohnAFR
writes:
>"JohnAFR" <joh...@aol.com> wrote in message
>news:ajknsi$lst$1...@bourbaki.localdomain
>
>> "Arny Krueger" <ar...@hotpop.com> wrote:
>
>>> I've measured unweighted vinyl SNR with a number of different
>>> test records, turntables, cartridges, tonearms, and
>>> preamplifiers. In no case have I observed vinyl playback with
>>> greater than 62 dB unweighted SNR. That gives vinyl resolution
>>> approximately equal to 10 bits.
>
>> That's very interesting. I'm curious how you measured SNR on
>> vinyl. Can you describe the procedure, test equipment, etc.
>> How did you determine the maximum signal level?
>
>The best tool I've found for measuring SNR is the relatively new High
>Fidelity News (HFN) test record. I establish maximum output level
>using its trackability test tracks. [clip]
Thanks, Arny, for the excellent explanation of you're measuring
process. Although, I believe your measurements are probably
very accurate, I feel they are a bit conservative due to the effect of
RIAA equalization, among other things. I think your results
demonstrate the need for weighted measurements when it comes
to the vinyl format. In other words, I believe that vinyl actually
performs more like a 12-bit system rather than a 10-bit system.
Thanks, again, for your detailed explanation.
Best regards,
John Elison
Arny Krueger
news:ajm1h6$k8p$1...@bourbaki.localdomain
> Hi Ears!
>
> Hmmm. SNR is not the end of the audio trail.
>
> It is possible to hear "into" the noise floor.
>
> sampling frequency of vinyl?
Since vinyl is extremely nonlinear and has severely compromised high
frequency dynamic range above 5-10 KHz, and digital does not, the
equivalence is not so easy to give. However it can be easily
demonstrated that digital sampling rates that are much above 32 KHz
have marginal benefits. You can find listening demonstrations related
to this at <http://www.pcabx.com/technical/low_pass/index.htm> and
<http://www.pcabx-pro.com/technical/sample_rates/index.htm> . Please
confine your judgments in this area to test results obtained under
Double Blind conditions (8 different DBT comparators for 4 different
operating systems/hardware platforms are now available at
www.pcabx.com ).
S888Wheel
>in any analog channel is proportional to the dynamic range of
>
>the medium times the bandwidth of the medium. The dynamic range
>is the ratio between the the maximum level encodable and the
>lowest level that is unambiguously separable from the noise
>
I am not very technically knowledgable on these things but it seems
like a tricky thing comparing the lowest level that is unambiguously
seperable from the noise on LP compared to CD. I was under the
impression that dithering was added noise used to help resolve low
level information on CDs. This seems counterintuitive based on the
criteria of how one determines bits of information. Do dithered CDs
have fewer bits of musical (signal) information than nondithered CD
counterparts?
Is it possible that gradiations of changes within the audable
dynamic range that are smaller than the lowest level that is
unambiguously seperable from noise may still have an audible affect
on the sound?
Richard D Pierce
TubeGarden <tubeg...@aol.com> wrote:
>Hi Ears!
>
>Hmmm. SNR is not the end of the audio trail.
>
>It is possible to hear "into" the noise floor.
Yes, in exactly the same way, for exactly the same reasons to
exactly the same degree one can also hear below the least
significant bit of properly digitized audio.
>And then the sampling frequency isuue, what did you take as the sampling
>frequency of vinyl?
The minimum sampling frequency needed is no less than twice the
highest useful frequency present on the disk. And anything
higher than that DOES NOT encode it with any greater accuracy.
Thus, given that most LPs are limited to 20 kHz and less
bandwodth, 40 kHz is the sample rate needed to completely and
faithfully encode it.
Arny Krueger
>> "JohnAFR" <joh...@aol.com> wrote in message
>> news:ajknsi$lst$1...@bourbaki.localdomain
>>> "Arny Krueger" <ar...@hotpop.com> wrote:
>>>> I've measured unweighted vinyl SNR with a number of different
>>>> test records, turntables, cartridges, tonearms, and
>>>> preamplifiers. In no case have I observed vinyl playback with
>>>> greater than 62 dB unweighted SNR. That gives vinyl resolution
>>>> approximately equal to 10 bits.
>>> That's very interesting. I'm curious how you measured SNR on
>>> vinyl. Can you describe the procedure, test equipment, etc.
>>> How did you determine the maximum signal level?
>> The best tool I've found for measuring SNR is the relatively new
>> High Fidelity News (HFN) test record. I establish maximum
>> output level using its trackability test tracks. [clip]
> Thanks, Arny, for the excellent explanation of you're measuring
> process. Although, I believe your measurements are probably
> very accurate, I feel they are a bit conservative due to the
> effect of RIAA equalization, among other things.
OK, "a bit" - throw in 6 dB one way or the other!
;-)
> I think your
> results demonstrate the need for weighted measurements when it
> comes to the vinyl format.
Thing is there is an equivalent to analog-domain weighting that
relates to the utilization of bits in the digital domain - called
"Noise Shaped Quantization". You can find a explanation of this and
a demo at <http://www.uni-jena.de/~pfk/mpp/dither.html> . To do the
listening demos in windows (they are IMO good) you need Winamp and
the .MPP plugin for Winamp. Unfortunately, the Winamp .MPP plugin
link that is linked out of that page (via the page heading) is
broken, but there's a good one at
http://orion.spaceports.com/~smaug/files.html .
>In other words, I believe that vinyl
> actually performs more like a 12-bit system rather than a 10-bit
> system.
If you noise-shape the quanitzation of 10 bits, you can probably
achieve sonic parity, as the "Noise Shaped Quantization" reference
will demonstrate right before your very ears.
Stewart Pinkerton
>>The amount of information
>>in any analog channel is proportional to the dynamic range of
>>
>>the medium times the bandwidth of the medium. The dynamic range
>>is the ratio between the the maximum level encodable and the
>>lowest level that is unambiguously separable from the noise
>>
>I am not very technically knowledgable on these things but it seems
>like a tricky thing comparing the lowest level that is unambiguously
>seperable from the noise on LP compared to CD. I was under the
>impression that dithering was added noise used to help resolve low
>level information on CDs.
It's added to decorrelate quantisation noise, thereby giving the
system a smooth noise floor with no quantisation artifacts. This has
the side effect of allowing the encoding and decoding of signals well
below the wideband noise floor.
>This seems counterintuitive based on the
>criteria of how one determines bits of information. Do dithered CDs
>have fewer bits of musical (signal) information than nondithered CD
>counterparts?
No, dither is roughly equivalent to 0.5 LSB, so a dithered CD will
still have 93dB of dynamic range, equal to 15.5 bits.
> Is it possible that gradiations of changes within the audable
>dynamic range that are smaller than the lowest level that is
>unambiguously seperable from noise may still have an audible affect
>on the sound?
Yes, it is possible to hear below the noise floor both of LP and CD,
that's a side effect of dither in the case of CD.
--
Stewart Pinkerton | Music is Art - Audio is Engineering
Richard D Pierce
S888Wheel <s888...@aol.com> wrote:
>>The amount of information
>>in any analog channel is proportional to the dynamic range of
>
>>
>>the medium times the bandwidth of the medium. The dynamic range
>>is the ratio between the the maximum level encodable and the
>>lowest level that is unambiguously separable from the noise
>>
>
>I am not very technically knowledgable on these things but it seems
>like a tricky thing comparing the lowest level that is unambiguously
>seperable from the noise on LP compared to CD. I was under the
>impression that dithering was added noise used to help resolve low
>level information on CDs.
What dithering does is several things: it decorrelates the
quantization artifacts from the signal, it does so essentially
by spreading the artifacts randomly thourgh the spectrum. It
linearizes the response of the system so that, averaged (as most
detectors, including the ear, operate), it does not have the
quantized response.
Neither of these, inherently, changes the amount of data in the
system. ONe might argue that the standard means of 1/2 LSB TPD
white dither reduces the drynamic range by 3 dB, but that is in
exchange for resolution far below the least significant bit.
That resolution is only achievable, as alluded to, because of
the properties of the detectors, the principle being the ear,
that's applied to these systems. Whether its an analog system
with noise or a dithered digital system, the reason we CAN here
below the floor is because the ear averages over time and
frequency both. Assuming you have a single tone buried in the
noise, say, at 4 kHz down 20 dB from the broadband (20 kHz)
noise floor. If the bandwidth of the detector is 20 kHz, you
won't find the signal, period.
It's because the ear is essentially a series of narrow-band
filters that this is possible. At 4 kHz, the width of one of
these filters is approximate 1/3 octave, from about 3.6 kHz to
4.5 kHz. That band is merely 1/22th of the total audio band, and
assuming the noise is evenly spread across that band (white),
that means that only 1/22 of the noise makes it through, while
ALL of the tone does. The result is that the noise in that band
is reduced to 1/22, or some 26 dB, while the signal is
unchanged. The signal, 20 dB BELOW the boroadband noise floor,
is now really 6 dB ABOVE the local noise once filtered, and is
not quite detectable.
>This seems counterintuitive based on the
>criteria of how one determines bits of information. Do dithered CDs
>have fewer bits of musical (signal) information than nondithered CD
>counterparts?
No, not inherently. In fact, it is arguable they have more.
> Is it possible that gradiations of changes within the audable
>dynamic range that are smaller than the lowest level that is
>unambiguously seperable from noise may still have an audible affect
>on the sound?
sure, and that why any system with a wider dynamic range, i.e.,
a lower level of ambiguation, MUST have higher resolution. The
two cannot be considered separately: they are precisely the
same, stated in different units. It is only in the high-end
audio business that they are considered different, and that's
simply one of many fundamental mistakes made in the high-end.
Arny Krueger
> I am not very technically knowledgeable on these things but it
> seems like a tricky thing comparing the lowest level that is
> unambiguously separable from the noise on LP compared to CD.
There is no trickery that I know of, and I love to find tricks.
> I
> was under the impression that dithering was added noise used to
> help resolve low level information on CDs.
That's included in the analysis that I've shown.
> This seems
> counterintuitive based on the criteria of how one determines bits
> of information.
Not really.
> Do dithered CDs have fewer bits of musical
> (signal) information than nondithered CD counterparts?
The math that has been shown relates to a correctly designed and
operational digital system. Dither is an inherent part of a properly
designed and operational digital system. Hopefully, there is no such
thing as an undithered CD counterpart, because that would be a CD
that is inherently defective.
> Is it possible that gradiations of changes within the audable
> dynamic range that are smaller than the lowest level that is
> unambiguously seperable from noise may still have an audible
> affect on the sound?
The inverse is true. It is possible to unambiguously separate signals
from noise by standard technical means when the signals are below
anything that is reliably audible by any known person. This is
routinely done.
Dave Collins
pat...@dircon.co.uk (Stewart Pinkerton) wrote:
> No, dither is roughly equivalent to 0.5 LSB, so a dithered CD will
> still have 93dB of dynamic range, equal to 15.5 bits.
I think most CD are dithered with TPDF or a variant, which is more
like 2 LSB's peak, but this doesn't really change the arguement....
What about the Stereophile article that showed LP's have all kinds
of response above 20k? Was it just 2nd harmonic distortion? I
can't remember if that possibility was mentioned in the text.....
DC
Espen Braathen
news:ajr4i...@enews2.newsguy.com...
> In article <ajp05t$877$1...@bourbaki.localdomain>,
> pat...@dircon.co.uk (Stewart Pinkerton) wrote:
>
> > No, dither is roughly equivalent to 0.5 LSB, so a dithered CD will
> > still have 93dB of dynamic range, equal to 15.5 bits.
>
> I think most CD are dithered with TPDF or a variant, which is more
> like 2 LSB's peak, but this doesn't really change the arguement....
TPDF dither reduces the noise free dynamic range with 4,8 dB.
Assuming a perfect chain, the dynamic range with 16 bit will be 93 dB
with dither.
DR = 6,0206n - 3 (dB)
--
Espen Braathen
WEB: http://espen-b.home.online.no
John Atkinson
> of response above 20k?
http://www.stereophile.com/showarchives.cgi?282:1 . As I describe in
the text, the >20kHz content was associated with cymbals and heavily
distorted electric guitar (and 3 clicks).
> Was it just 2nd harmonic distortion? I can't remember if that
> possibility was mentioned in the text.....
It wasn't, but I did wonder about that at the time, as well as
mistracking. The phono cartridge used had quite low THD when I
measured it a long while ago; certainly it didn't sound like it
was mistracking at the times when the spectrum had ultrasonic
content and things like hi-hat cymbal were not recorded that "hot."
John Atkinson
Editor, Stereophile
Mark Ovchain
> Hmmm. SNR is not the end of the audio trail.
Certainly not! One of the few people I respect who used to nose
around here called it, I think, "mostly useless".
> It is possible to hear "into" the noise floor.
Dear me, what's that got to do with anything? You can hear into a
noise floor, be it analog, digital, hybrid, or anything else. Again,
several people here have provided clear explainations of just how the
auditory system performs this simple and classical signal-detection
operation.
> And then the sampling frequency isuue, what did you take as the sampling
> frequency of vinyl?
Now why would you have to choose any sampling frequency? According
to Shannon, you only have to integrate the log base 2 of the SNR
over the total bandwidth to get the actual "bits" available in any
signal. You don't have to even consider sampling.
Now, you get into a couple of different problems with LP's in that
regard, because you effectively have only 1 channel at higher
frequencies (25dB is more than enough for the ear, only 4 bits/Hz in
terms of actual bitrate, though), you have a peak energy falloff at
high and low frequencies (which drops the SNR by dropping the
'S'ignal level), you have distortion products that count at high
levels as part of the "noise", and so on, which cuts the raw SNR of
60dB or so for a good table far, far down.
Also, it's not fair counting the 100kHz artifacts of mistracking and
groove dynamics of an LP, they aren't part of the information, only
part of the distortion products, so the bandwidth of LP really isn't
100K, it's maybe 40K if you're doing well, and you have new
equipment, a new stylus, and a new groove to deal with. While this
might seem like an advantage for LP, you also have fewer bits, less
channel separation, and so on, results that really do tend to
seriously impair the use of LP as a source of bits.
So I'm pretty sure that Redbook CD wins, although not by a landslide.
Mark Ovchain
> I am not very technically knowledgable on these things but it seems
> like a tricky thing comparing the lowest level that is unambiguously
> seperable from the noise on LP compared to CD. I was under the
> impression that dithering was added noise used to help resolve low
> level information on CDs. This seems counterintuitive based on the
> criteria of how one determines bits of information. Do dithered CDs
> have fewer bits of musical (signal) information than nondithered CD
> counterparts?
A fair question, indeed.
The answer is that a digitally sampled signal that is properly
dithered contains rather more of the original information than one
that was not properly dithered.
The reason is simple. Check out some of Bart Locanthi Sr's papers in
the Journal of the Audio Engineering Society, or the original works
on dithering that date back to the 1950's in the telecom industry.
Mark Ovchain
Oh dear, I'll have to partially disagree here, I fear.
> What dithering does is several things: it decorrelates the
> quantization artifacts from the signal, it does so essentially
> by spreading the artifacts randomly thourgh the spectrum.
Agreed.
> It
> linearizes the response of the system so that, averaged (as most
> detectors, including the ear, operate), it does not have the
> quantized response.
Agreed again.
> Neither of these, inherently, changes the amount of data in the
> system.
Well, yes, but in a misleading sense, I fear. The total data
in the system is fixed by bits*sampling rate, however, slighly
more information, on the average, about the original signal
is transmitted when the signal is properly dithered.
Consider the case of the poor sine wave at +-.4 LSB. Without
dithering, it disappears. With dithering, it is present, in the
noise, but generally quite detectable.
> ONe might argue that the standard means of 1/2 LSB TPD
> white dither reduces the drynamic range by 3 dB, but that is in
> exchange for resolution far below the least significant bit.
Agreed, or fairly so.
> Whether its an analog system
> with noise or a dithered digital system, the reason we CAN here
> below the floor is because the ear averages over time and
> frequency both.
But wouldn't you call the particular kind of averaging
(in the case of the ear) filtering?
> Assuming you have a single tone buried in the
> noise, say, at 4 kHz down 20 dB from the broadband (20 kHz)
> noise floor. If the bandwidth of the detector is 20 kHz, you
> won't find the signal, period.
I beg to differ, sir, the right kind of correlation detector can
in fact find that, but you may argue, even though the input bandwidth
of the detector is broadband, that it does so by creating a kind
of filter, so I will have to say that perhaps this depends on
use of language.
> >This seems counterintuitive based on the
> >criteria of how one determines bits of information. Do dithered CDs
> >have fewer bits of musical (signal) information than nondithered CD
> >counterparts?
> No, not inherently. In fact, it is arguable they have more.
Indeed, said this way, I'm in complete agreement. I'm simply afraid
that somebody will take your comment above in a noxious fashion.
Richard D Pierce
As a point of serious curiosity, how would one know, by
listening, whether there were ultrasonic artifacts of
mistracking high-frequency information if the mistracking had,
say, minimal or non-existant manifestations in the audio band?
(There's a simple answer that's only partially true...).
One of the things that I noticed from tests I had done some time
ago is that because of the spectrum of such things as cymbals is
so broad and dense, it sometimes takes staggering amounts of
non-linearity to become "audibly" distorted. A pathological
example is what does white noise with 20% THD sound like
(answer: it depends: if it's all clipping, it's audible, other
forms are completely inaudible).
I think the kinds of test mentioned are far from definitive. I
might like to see, for example, the same material cut at several
different levels, and then do relative comparisons of the amount
of high-frequency information present. This might at least give
a suggestion of how much is real data vs distortion artifacts. I
would also like to see a comparison between the original master
and the LP.
My own informal experiments would indicate that mistracking is
more of an issue than assumed, in that the amount of >20 kHz
information is VERY level dependent in ways that does not appear
in other media or in direct instrumental measurements.
In any case, the number of such instances where ultrasonic
information could be real are exceedingly rare, in my
expreience. And, given my well-known eclectic musical
preferences, are almost ALWAYS associated with albums of
Exceeding Musical Boredom(tm). :-)
Dave Collins
stereophi...@compuserve.com (John Atkinson) wrote:
> You can find the specific instance of the LP's HF content at
> http://www.stereophile.com/showarchives.cgi?282:1 . As I describe in
> the text, the >20kHz content was associated with cymbals and heavily
> distorted electric guitar (and 3 clicks).
Aren't these exactly the type of signals that might cause mistracking?
I know a little about disk cutting (by no means an expert) and would
have to venture that the >15k stuff IS just distortion.
Unless we are arguing that the LP has infinite bandwidth and
resolution, that is......
DC
TubeGarden
>have to venture that the >15k stuff IS just distortion.
>
>Unless we are arguing that the LP has infinite bandwidth and
>resolution, that is......
Hi Ears!
Q: How many angels can dance on the head of an audiophile?
A: None. They dance in the head of an audiophile.
LPs, even those made from digital masters, are analog surfaces. Bits
is in the mind of the beholder, not on the LP.
Music cannot be recorded, it can only be sketched for later imagining
by a listener.
Reality may or may not be shared between the musicians and the
listener. Never certain.
This accuracy stuff is a shell game. Easy to concieve, but, boring -
without music.
We each hear alone. God knows what :)
Happy Ears!
Al
Alan J. Marcy
Phoenix, AZ
PWC/mystic/Earhead
Dave Collins
tubeg...@aol.com (TubeGarden) wrote:
> LPs, even those made from digital masters, are analog surfaces. Bits
> is in the mind of the beholder, not on the LP.
Didn't I once hear that a digitally mastered LP could damage the
bearing of my turntable? I think there might be a class action
lawsuit there!
DC
Cannot Understand Why You Dis AOL
I read someone (Mark Levinson, IIRC) say that *listening* to digital
recordings *is physically harmful to the listener*.
He was supposed to make a big annoncement withe evidence to the
effect...two years ago. Still waiting.
_______
-S.
Cannot Understand Why You Dis AOL
>>I know a little about disk cutting (by no means an expert) and would
>>have to venture that the >15k stuff IS just distortion.
>>
>>Unless we are arguing that the LP has infinite bandwidth and
>>resolution, that is......
>
>Hi Ears!
>
>Q: How many angels can dance on the head of an audiophile?
>
>A: None. They dance in the head of an audiophile.
>
>LPs, even those made from digital masters, are analog surfaces. Bits
>is in the mind of the beholder, not on the LP.
and what is the playing surface of a CD?
or of a metal oxide tape?
>Music cannot be recorded, it can only be sketched for later imagining
>by a listener.
pass the bong.
_______
-S.
Richard D Pierce
This is another example of the "Diamond" effect. Dr. J. Diamond
gave a talk before the 64th AES convention titled: "Human
Stress Provoked by Digitalized Recordings," in which he claimed:
"volunteers had their arms pushed down more easily by the
speaker during the digitally recorded passages"
Diamond's techniques were roundly criticized for being sloppy,
unscientific and unreliable. Diamond's choice was to continue by
stating (in db magazine, 1981 Jan) that his method:
"promises results for nearly any practitioner"
and that the effects are profound and easily detected.
Attempts were made to duplicate his results, but to no avail.
Not only were Diamond's inexact "arm pushing tests" attempted,
but other, more widely accepted measures of stress and muscle
tone utiltized. In one test, the tester was even coached by Dr.
Diamond as to the exact technique. And the results were
negative, failing altogether to reproduce Diamond's results.
Diamond's response (JAES 1981 Oct) was little more than a
collection of caveats and excuses:
"There is a level of expertise required to carry out the test
successfully."
contrary to his earlier promise of results for nearly any
practitioner. Further,
"I have found, for reasons of which I am unaware, that
increasing the length of lead between pickup and preamplifier,
or preamplifier and amplifier, would have the effect, under
many circumstances, of overcoming the digital effect.
Really?
"parameters such as EEG and GSR did not reveal the problem. Of
this I am well aware. To date the problem is revealed only by
the test I advocate."
and
"Only when they attend to all the variables as I do in
my research, and when they develop a level of expertise
in performing the test, will it be possible to compare
their results with mine."
again, contrary to his first claim.
In an extraordinary statement, he claims:
"On an average day I perform the test 350 to 500 times, and
I have done this for many years. This is certainly more than
the eleven or so "push tests."
This claim is extraordinary on several fronts:
1. Assume Dr. Diamond works an 8 hour day. 500 tests in an 8
hour period mean he administers the test on average 63 times
per hour, or about once per minute. When does Dr. DIamond
attend to other patient needs? Paperwork? Going to the
bathroom. The physicians I talk to suggest the notion that
such a test, applied in a meaningful fashion sufficient to
derive good data 500 times a day is, to ut it mildly, a
bit hyperbolic
2. The "eleven or so push tests" he criticizes is precisely the
method he used in his original talk.
Diamond's claims have been thoroughly discredited, and it is
Diamond himself that's responsible for it: his sloppiness,
back-pedaling, excuses and contradictions do lend credibility to
his claim.
And there are thus who would argue that Mark Levinson isn't half
the careful, skilled researcher that Dr. Diamond is.
Yes, we await Levinson's data, but we are hardly holding our
breath.
Dave Collins
stev...@aol.com (Cannot Understand Why You Dis AOL) wrote:
> I read someone (Mark Levinson, IIRC) say that *listening* to digital
> recordings *is physically harmful to the listener*.
Actually what they claim is that certain types of digital coding
(i.e. PCM) is bad for your health, while DSD has no effect. There's a
guy called Dr. Diamond that performs a test of arm strength before
and after playing a CD and, hey presto, PCM is evil....
DC
Norman M. Schwartz
news:ak3oq...@enews1.newsguy.com...
> Dave Collins dcol...@earthlink.net wrote:
> >
> >Didn't I once hear that a digitally mastered LP could damage the
> >bearing of my turntable? I think there might be a class action
> >lawsuit there!
If you have access to a 20+ year old collection of *TAS* you can see
an electron micrograph of a TT bearing which was in fact damaged by
playing a digitally recorded LP!. If I hadn't put my TAS collection
in its proper place, specifically curbside for the sanitation
engineers, I'd provide the Volume and page number reference. (Perhaps
some older audiofools can provide it.)
Dave Collins
"Norman M. Schwartz" <nm...@att.net> wrote:
> If you have access to a 20+ year old collection of *TAS* you can see
> an electron micrograph of a TT bearing which was in fact damaged by
> playing a digitally recorded LP!.
So what does "in fact" mean in this context?
DC
Richard D Pierce
"In fact" means that Harry Pearson or one of his writers made it
up, and thus it is fact. In a similar vein, another article
proved that "in fact" is is impossible for CD's to have 16 bots
of resolution. The "though process" went something like this:
First, it has a sign bit, so that means it can only be 15 bits.
Next, 16 bits is made out of two bytes,m and each byte must have
a parity bit, so take away two more bits, leaving 13 bits,
etc...
The great thing about absolute sound "in fact" is that they
never let little inconveniences like "facts" get in their way.
The technical grasp of many of that magazine's authors would be
laughable in the extreme were it not for the fact that they and
a significant portion of their readership takes them so
seriously.
Bob Olhsson
<DPi...@TheWorld.com> wrote:
>This claim is extraordinary on several fronts:
The MOST extraordinary being that the same observations were made
despite the fact that, unknown to Dr. Diamond, the sound system at that
AES meeting was being digitally processed!
--
Bob Olhsson Audio Mastery Recording Project Design and Consulting
Box 90412, Nashville TN 37209 Tracking, Mixing and Mastering
615.352.7635 FAX 615.356.2483 Mix Evaluation and Quality Control
40 years of making people sound better than they thought possible!