28 Jul
2003
28 Jul
'03
6:26 a.m.
I'm a professional audio engineer for 15 years, and DSD did convince me. I
could compare the sound quality with TI ADC/DAC.
Conventional PCM techniques are unable to reproduce high frequencies
correctly. And the explanation is very simple. If you record a sound at 44.1
kss, you get a theorical frequency response of 0 - 22050 Hz. BUT to describe
frequencies from 11050 to 22050 Hz, you can only play with a 4-sample long
period.
A 22050 Hz sine could be really accurate (one sample up, one sample down
every 1/22050th second), and so is 11025. But intermediary frequencies
introduces temporal aliasing, some metallic feeling due to temporal
quantization. This is inherent to the very low sampling rate (96 kHz is just
a bit better, but no miracle), which is unable to describe waveforms at high
frequencies.
Bad high frequencies temporal definition means bad transients. Anyone can
notice it when he _actually_ hear and compare PCM and DSD.
Stop speculative talking and try to get some real demo...
--
mickael
28 Jul
28 Jul
7:17 a.m.
New subject: [linux-audio-dev] Direct Stream Digital / Pulse Density Modulation musing/questions
Hello Michael,
At 10.13 28/07/2003 +0200, you wrote:
...
A 22050 Hz sine could be really accurate (one sample
up, one sample down
every 1/22050th second), and so is 11025. But intermediary frequencies
introduces temporal aliasing, some metallic feeling due to temporal
quantization. This is inherent to the very low sampling rate (96 kHz is just
...
sorry, but this is incorrect. The sampled signal has nothing to do with the
information it carries. The sampled signal just contains the information
needed to get back the original signal but it isn't the original signal
(despite at lower frequency it pretty much resemble the original signal),
you have to pass it through a reconstruction filter to get the original
(bandlimited) signal back. You can't simply look at the sampled signal to
see what the original signal was, much the same way you can't just look at
a compressed file to see what the original file was. This is one of the
most common misconception circulating about PCM, but it is indeed a
misconception.
If you hear a difference between DSD and PCM the problem should be
somewhere else, may be in the ADC/DAC used, may be in the DSD itself, but
sure not in what you describe above.
Bye,
--
Denis Sbragion
InfoTecna
Tel: +39 0362 805396, Fax: +39 0362 805404
URL: http://www.infotecna.it
10:14 a.m.
New subject: [linux-audio-dev] Direct Stream Digital / Pulse Density Modulation musing/questions
It becomes even more interesting when you bring in dither. Then you can
represent signals whose amplitude is less than one bit, and you can
increase time resolution to infinity.
Also, DSD is so cumbersome to process, that it is often converted to and
from PCM during production to be able to apply digital effects and
stuff like that. So it is very likely that an SACD you buy has been
through PCM somewhere in the production.
Audibility differences between DSD and PCM are more or less meaningless
to discuss, it is next to impossible to make a correct comparison, and
differences are so small that almost no-one is able to hear them. Most
cannot reliably hear a difference between 44.1 and 96 kHz. I know I
can't. Increasing the dynamic range from 16 to 24 bits is a much
greater win than increasing the sample rate.
Sony and others has put out straight lies to the public in the
commercials where they compare DSD and PCM, which is one reason why
there are so many myths about DSD vs PCM around.
/Anders Torger
On Monday 28 July 2003 10.58, Denis Sbragion wrote:
Hello Michael,
At 10.13 28/07/2003 +0200, you wrote:
...
A 22050 Hz sine could be really accurate (one
sample up, one sample
down every 1/22050th second), and so is 11025. But intermediary
frequencies introduces temporal aliasing, some metallic feeling due
to temporal quantization. This is inherent to the very low sampling
rate (96 kHz is just
...
sorry, but this is incorrect. The sampled signal has nothing to do
with the information it carries. The sampled signal just contains the
information needed to get back the original signal but it isn't the
original signal (despite at lower frequency it pretty much resemble
the original signal), you have to pass it through a reconstruction
filter to get the original (bandlimited) signal back. You can't
simply look at the sampled signal to see what the original signal
was, much the same way you can't just look at a compressed file to
see what the original file was. This is one of the most common
misconception circulating about PCM, but it is indeed a
misconception.
If you hear a difference between DSD and PCM the problem
should be somewhere else, may be in the ADC/DAC used, may be in the
DSD itself, but sure not in what you describe above.
Bye,
--
Denis Sbragion
InfoTecna
Tel: +39 0362 805396, Fax: +39 0362 805404
URL: http://www.infotecna.it 
11:19 a.m.
New subject: [linux-audio-dev] Direct Stream Digital / Pulse Density Modulation musing/questions
On Mon, Jul 28, 2003 at 02:00:41PM +0200, Anders Torger wrote:
It becomes even more interesting when you bring in
dither. Then you can
represent signals whose amplitude is less than one bit, and you can
increase time resolution to infinity.
Time (phase) resolution *is* already infinite, even without dither.
The quantisation of time and amplitude are two completely different beasts.
--
FA
12:15 p.m.
New subject: [linux-audio-dev] Direct Stream Digital / Pulse Density Modulation musing/questions
On Monday 28 July 2003 14.59, Alfons Adriaensen wrote:
On Mon, Jul 28, 2003 at 02:00:41PM +0200, Anders
Torger wrote:
Well, not exactly. I'm am by no means an expert on dither, so if I have
wrong, let me know.
Think that you are sampling a sine wave, without dither. Truncating to
an even sample can result in a small phase shift after the
reconstruction, which is not there if you make a properly dithered
recording. Thus - dither improves temporal resolution, through
improvement of amplitude resolution.
Am I right?
/Anders Torger
It becomes even more interesting when you bring
in dither. Then you
can represent signals whose amplitude is less than one bit, and you
can increase time resolution to infinity.
Time (phase) resolution *is* already infinite, even without dither.
The quantisation of time and amplitude are two completely different
beasts. 
12:44 p.m.
New subject: [linux-audio-dev] Direct Stream Digital / Pulse Density Modulation musing/questions
Anders Torger wrote:
On Monday 28 July 2003 14.59, Alfons Adriaensen
wrote:
Sounds spot-on from here...
It's important to remember that dithering "works" through averaging, and as
such can only provide a finite level of benefit. I do not believe there is
such a thing as infinite phase resolution with PCM... only effective phase
resolution, which varies depending on the dither, signal content and length
being analysed.
See how many different answers you can get from the FFT of a 997Hz sine wave
dithered to 16-bit and generated at 44100Hz by changing the window length
and type... :)
- Tom.
On Mon, Jul 28, 2003 at 02:00:41PM +0200, Anders
Torger wrote:
Well, not exactly. I'm am by no means an expert on dither, so if I
have wrong, let me know.
Think that you are sampling a sine wave, without dither. Truncating to
an even sample can result in a small phase shift after the
reconstruction, which is not there if you make a properly dithered
recording. Thus - dither improves temporal resolution, through
improvement of amplitude resolution.
Am I right? It becomes even more interesting when you bring
in dither. Then you
can represent signals whose amplitude is less than one bit, and you
can increase time resolution to infinity.
Time (phase) resolution *is* already infinite, even without dither.
The quantisation of time and amplitude are two completely different
beasts.
1:29 p.m.
New subject: [linux-audio-dev] Direct Stream Digital / Pulse Density Modulation musing/questions
On Mon, Jul 28, 2003 at 04:01:31PM +0200, Anders Torger wrote:
On Monday 28 July 2003 14.59, Alfons Adriaensen
wrote:
Yes, in that sence you are absolutely right. The point I wanted to make
is that sampling by itself does not introduce any quantisation of the
time axis - in the reconstructed signal, the places that correspond to
the sample points are not 'better' than the parts in between.
Some people *do* believe such things...
--
FA
On Mon, Jul 28, 2003 at 02:00:41PM +0200, Anders
Torger wrote:
Well, not exactly. I'm am by no means an expert on dither, so if I have
wrong, let me know.
Think that you are sampling a sine wave, without dither. Truncating to
an even sample can result in a small phase shift after the
reconstruction, which is not there if you make a properly dithered
recording. Thus - dither improves temporal resolution, through
improvement of amplitude resolution.
Am I right? It becomes even more interesting when you bring
in dither. Then you
can represent signals whose amplitude is less than one bit, and you
can increase time resolution to infinity.
Time (phase) resolution *is* already infinite, even without dither.
The quantisation of time and amplitude are two completely different
beasts.
12:06 p.m.
New subject: [linux-audio-dev] Direct Stream Digital / Pulse Density Modulation musing/questions
Hello Anders,
At 14.00 28/07/2003 +0200, you wrote:
It becomes even more interesting when you bring in
dither. Then you can
represent signals whose amplitude is less than one bit, and you can
increase time resolution to infinity.
...
well I agree with anything else you have written but here there's a basic
relation that sets a limit on time resolution of a linear system:
df * dt > 1/2
Here d is "delta" and not "differential", f is frequency and t is
time.
This relation applies to any signal and/or any linear system. So the
limited bandwith of a PCM system sets a lower limit on time resolution. For
sure dithering allow for the almost complete use of the available time
resolution, and also increasing the semple rate increase time resolution,
but as usual the big question is: does it matters to our ear? I think no,
and with this respect I completely agree with you.
Bye,
P.S. I have yet to test the latest BF with fftw3. My RedHat 7.3 has a
bugged gcc which fails to compile the sse support for fftw3, losing much of
the advantages of fftw3. I filed a bug report and I'm waiting for an answer
from RedHat.
--
Denis Sbragion
InfoTecna
Tel: +39 0362 805396, Fax: +39 0362 805404
URL: http://www.infotecna.it
8:45 a.m.
New subject: [linux-audio-dev] Direct Stream Digital / Pulse Density Modulation musing/questions
[...]
Conventional PCM techniques are unable to reproduce
high frequencies
correctly. And the explanation is very simple.
Actually a correct explanation isn't that simple. Yours is much _too_
simple.
Theoretically a 20 kHz bandlimited signal can be represented _exactly_ as a
40 kHz PCM stream. In order to not have to use a very steep lowpass filter
in
the DAC it is better to use a somewhat higher sampling frequency. 48 kHz
should
be enough most of the time.
If you record a sound at 44.1
kss, you get a theorical frequency response of 0 - 22050 Hz. BUT to
describe
frequencies from 11050 to 22050 Hz, you can only play
with a 4-sample long
period.
These are interpolated using a lowpass filter.
A 22050 Hz sine could be really accurate (one sample
up, one sample down
every 1/22050th second), and so is 11025. But intermediary frequencies
introduces temporal aliasing, some metallic feeling due to temporal
quantization. This is inherent to the very low sampling rate (96 kHz is
just
a bit better, but no miracle), which is unable to
describe waveforms at
high
frequencies.
Bad high frequencies temporal definition means bad transients. Anyone can
notice it when he _actually_ hear and compare PCM and DSD.
Stop speculative talking and try to get some real demo...
This has been done and it is very hard if not impossible to hear the
difference.
24bit 96 kHz is most likely better than DSD, _especially_ at high
frequencies.
The demos for SACD often compare normal stereo PCM to surround on a
SACD...
Actually, DSD suffers deteriorating sample resolution
at higher frequencies whereas the resolution doesn't depend on the frequency
for
PCM. Also signal processing will have to be done in PCM. A SACD player with
an EQ has DSD -> PCM and PCM -> DSD convertors and these convertions
are not lossles.
The DACs and monitors you use probably make a lot more difference than
44 kHz PCM or 96 Khz or DSD.
There are also people who buy very expensive oxygen free cables and claim
to hear the difference. There are thousands of religious people who _know_
God exists...
DSD is a bad thing IMHO.
--ms
7817
days inactive
7817
days old
linux-audio-dev@lists.linuxaudio.org
8 comments
6 participants
participants (6)
-
Alfons Adriaensen -
Anders Torger -
Denis Sbragion -
Martijn Sipkema -
Mickael S Vardo -
Tom Browne