"William Sommerwerck" wrote in message
...
But didn't those recordings all start out as PCM ?
No.
Yes.

In other words, DSD recordings are really PCM? Since when?


I the case of digital recording, from the original AD stage at least, and
probably thru a DAW or other digital multitrack system, and maybe even thru
the mastering stage.

Obviously doesn't apply to analogue recordings that are made or remastered
to exclusively DSD gear, if there is such a thing as a DSD A-D converter ...

geoff

Geoff@work wrote:
"Randy Yates" wrote in message
...

Yeah, I guess Dr. Dolittle wasn't around when Sony admitted to the AES
that
all SACD recordings start out as a kind of PCM.
While that may have been true several years ago, I believe Glen
Zelnicker (Z-Systems) now has a DAW that allows the signal to orginate
in and remain DSD throughout the mixing chain.

Which will apply to *some* new recordings.

Maybe there should be a new code a la ADD, like A-D-DSD, or D-D-DSD, or
D-DSD-DSD, etc.

Aren't most PCM converters today just 1-bit delta-sigma run into a
counter that is periodically sampled? If so, it's not clear to me what
the difference really is between that and continuing on with just the
1-bit DSD stream throughout.

I know darned well modern audio converters aren't flash and I don't
think they are even successive approximation any more.


Bob
--

"Things should be described as simply as possible, but no simpler."

A. Einstein

Bob Cain wrote:



Dirk Bruere at Neopax wrote:

Bob Cain wrote:



Dirk Bruere at Neopax wrote:

Well, we have some new types of speaker that really are
revolutionary (to the extent that I have not seen anything like them
on the market). Doing the frequency equalisation across the spectrum
using onboard DSPs is part of what I'm working on.



Who's "we"?


'We' - well, that's a rather interesting question.
Does the name 'Paul Dobson' ring any bells here?


It rings a bell but I can't place it.

Founder of StudioMaster

Is this a company effort? If so I was wondering what company it might be.

Startup, so new we do not yet have a registerd name.

I'm glad to see this application maturing. It's what got me into DSP 35
years ago before it was called DSP. Alas, back then the technology was
far too expensive, not to mention bulky, to do the conversions or to
implement the characterization and inversion algorithms I found (as in
located, not created) and coded. The work ended up just being a
feasibility study with the conclusion that it was doable but not right
then at a commercial level.

I, for one, won't be the least surprised if your results are stunning.

It's not so much the DSP, but the configuration of the multiple speakers.
Can't really tell you much more until we go public.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org

"Dirk Bruere at Neopax"
Phil Allison wrote:


** How smartarse.

You are bereft of even the tiniest clue.

The hi-fi audio charlatans & scammers will LOVE you.

Love me, hate me... I don't care as long as they pay me.


** Better get the dough up front.

If they chose to pay on "results", you are ****ed.

Let me put it this way...


** I'm not the least interested in old as the hills BULL CRAP like that
below.

No need for YOU to worry about maybe being ripped off by a hi-fi charlatan.

It is a guaranteed certainty.



.......... Phil




We had one guy walk in, listen to some prototypes and immediately asked to
be put at the top of the list when they are sold. They were not even fully
equalised across the spectrum at the time, just hand tweaked on the
bass/treble knobs of the driving amp.

They *will* sell - I have no doubt about that at all. Everyone who hears
them is really impressed. Plus, they look amazing.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org

Phil Allison wrote:

"Dirk Bruere at Neopax"

Phil Allison wrote:


** How smartarse.

You are bereft of even the tiniest clue.

The hi-fi audio charlatans & scammers will LOVE you.

Love me, hate me... I don't care as long as they pay me.


** Better get the dough up front.

If they chose to pay on "results", you are ****ed.

Let me put it this way...



** I'm not the least interested in old as the hills BULL CRAP like that
below.

No need for YOU to worry about maybe being ripped off by a hi-fi charlatan.

It is a guaranteed certainty.

Er... I'm not buying - I'm selling.
Remember?

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org

"Dork Breuer at Neolax"


** How smatters.

You are bereft of even the tiniest clue.

The hi-if audio charlatans & scammers will LOVE you.

Love me, hate me... I don't care as long as they pay me.


** Better get the dough up front.

If they chose to pay on "results", you are ****ed.

Let me put it this way...



** I'm not the least interested in old as the hills BULL CRAP like
that below.

No need for YOU to worry about maybe being ripped off by a hi-if
charlatan.

It is a guaranteed certainty.

Err... I'm not buying - I'm selling.
Remember?



** You really are a colossal ****ing ass - Dork !!!!!!

The issue is whether YOU will get paid for your work.


Quote :

" Love me, hate me... I don't care as long as they pay me."



........ Phil

Phil Allison wrote:

"Dork Breuer at Neolax"

** How smatters.

You are bereft of even the tiniest clue.

The hi-if audio charlatans & scammers will LOVE you.

Love me, hate me... I don't care as long as they pay me.


** Better get the dough up front.

If they chose to pay on "results", you are ****ed.

Let me put it this way...



** I'm not the least interested in old as the hills BULL CRAP like
that below.

No need for YOU to worry about maybe being ripped off by a hi-if
charlatan.

It is a guaranteed certainty.

Err... I'm not buying - I'm selling.
Remember?




** You really are a colossal ****ing ass - Dork !!!!!!

The issue is whether YOU will get paid for your work.

Er... I'm one of the three partners in the startup.


Quote :

" Love me, hate me... I don't care as long as they pay me."

Don't worry - I'll keep you informed.
BTW, you wasted your money on that anger management course.
Better luck with your next psychiatrist.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org

"Dirk Bruere at Neopax" schrieb im Newsbeitrag
...

So what advantage does this offer over, say, some .wav 32 bit floating
point sampled at 384 kHz?

Technically or in sound?

The sound is simply undistorted. Everything becomes volume, weight,
timbre, which are nornally loosen by recording. Musicians and singers
sounding like adult men and women. The differences are in the normal
human frequency and dynamic range, without hokus-pokus.

The technic allows a realistic view on digital audio. Many problems,
symptomes can be judjed as self made. If you avoid them to raise, you
must not fight them. The key word is switching distortion or glitches.
The ones the raising everywhere and always by switching from 0 to 1 or
back.

An example: Digital signal is aritmic, not systematic impulses. Two or
three contigous 1-s or 0-s should mean DC current for that time. The
overswinging in a transistor after the switching rings out. The power
supply can answer the next request. This was the ground to develop the
EFM technic, by production of optical discs. Well, if you now make an
128x oversampling, then it is not DC current, it is a burst for the same
time range. 128 times the non-rithmic request to the power supply too.
The signal will be covered due a veil from glitches, they cannot ring
out anymore. The distortion is there in the whole frequency and dynamic
range. The interaction between the electronic and power supply brings
the next conflict: The irregular bursts causing power requests which
will be not fulfilled in the right time. The result is growing jitter.
And so on.

So now we have a better understanding and handling to things like
pre-ringing of low-pass filters, accumulation of quantisation noise and
quantisation distortion, the sense of high sample rate, high bit rate,
oversampling, upsampling, noise shaping, dithering and so on. The whole
scientific background with the research results takes several hundred
pages. We are working on a simplified Flash-animated "paper", to answer
the questiions of our customers too. I hope it wil be ready soon.

Best regards
--
Johann Spischak

SDG, Spischak Digital GmbH
+49-911-965-7319
http://sdg-master.com

"Dork Bruere at Neopax"


** You really are a colossal ****ing ass - Dork !!!!!!

The issue is whether YOU will get paid for your work.

Er... I'm one of the three partners in the startup.


** Then it was** YOUR** mistake to post this ****e :


" Love me, hate me... I don't care as long as they pay me."


** Partners in CRIME is all you are.

Being the dumb, autistic one means Dork will be the big loser.

LOL



........ Phil

Phil Allison wrote:

"Dork Bruere at Neopax"


** You really are a colossal ****ing ass - Dork !!!!!!

The issue is whether YOU will get paid for your work.

Er... I'm one of the three partners in the startup.



** Then it was** YOUR** mistake to post this ****e :


" Love me, hate me... I don't care as long as they pay me."

I was referring to the customers.


** Partners in CRIME is all you are.

Being the dumb, autistic one means Dork will be the big loser.

LOL

I think you are probably mentally ill.


--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org

Dirk Bruere at Neopax wrote:
Phil Allison wrote:

[P.A.'s usual trash snipped]


I think you are probably mentally ill.

What was your clue? :-)


Bob
--

"Things should be described as simply as possible, but no simpler."

A. Einstein

"Dork Bruere at Neopax"

= biggest FOOL imaginable



** You really are a colossal ****ing ass - Dork !!!!!!

The issue is whether YOU will get paid for your work.

Er... I'm one of the three partners in the startup.



** Then it was** YOUR** mistake to post this ****e :


" Love me, hate me... I don't care as long as they pay me."

I was referring to the customers.


** Then you are an ever BIGGER ****ing ass !!


" ** The hi-fi audio charlatans & scammers will LOVE you."

Love me, hate me... I don't care as long as they pay me. "


The "charlatans and scammers" above are **NOT** the dupes .


YOU COLOSSAL ****ING IDIOT !!!!





......... Phil

"Boob Cain"


= a congenital, autistic pig ignorant asshole.




......... Phil

Phil Allison wrote:

"Boob Cain"


= a congenital, autistic pig ignorant asshole.




........ Phil

I know of several people over in sci.physics like you.
Always foaming at the mouth whenever Einstein and his 'obviously fake' theories
are mentioned. They too substitute invective for substance and are equally
beyond reason.

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org

Dirk Bruere at Neopax wrote:

If they chose to pay on "results", you are ****ed.

Let me put it this way...
We had one guy walk in, listen to some prototypes and immediately asked to be
put at the top of the list when they are sold. They were not even fully
equalised across the spectrum at the time, just hand tweaked on the bass/treble
knobs of the driving amp.

You mean Dobson's ultra-heavy bass boost ?

They *will* sell - I have no doubt about that at all. Everyone who hears them is
really impressed. Plus, they look amazing.

I know, I designed them.

Look where I am now !

Graham

"Dork Bruere at Neopax"



** Poor Dork,

it must be depressing for him being born a congenital idiot.






......... Phil

An example: Digital signal is aritmic, not systematic impulses. Two
or three contigous 1-s or 0-s should mean DC current for that time.
The overswinging in a transistor after the switching rings out. The
power supply can answer the next request. This was the ground
to develop the EFM technic [sic], by production of optical discs.

EFM is (eight-to-fourteen modulation) is a method to increase data
robustness. It has nothing to do with controlling the load on the power
supply.

"Geoff@work" writes:
[...]
Obviously doesn't apply to analogue recordings that are made or remastered
to exclusively DSD gear, if there is such a thing as a DSD A-D converter ...

That would be a delta-sigma A/D minus the decimator. I haven't seen one, though,
but it should be easy to make.
--
% Randy Yates % "So now it's getting late,
%% Fuquay-Varina, NC % and those who hesitate
%%% 919-577-9882 % got no one..."
%%%% % 'Waterfall', *Face The Music*, ELO
http://home.earthlink.net/~yatescr

"Chel van Gennip" schrieb im Newsbeitrag
...
On Thu, 19 Jan 2006 03:33:23 +0100, Johann Spischak wrote:

Nice to just mention a few words without any indication of the real
life
effect. One example: "quantisation noise" the magnitude of this error
is
about 1/2 LSB or -150dB below FS for 24 bit sampling. That is a level
where even analog circuits suffer from quantum effects.

You are mixing two different things. Quantisation noise roots from the
ringing of the low pass filter in the ADC, raises according and
depending on the loudness. It is in the hearable dynamic range. Against
this was noise shaping born. Accumulation of this, like in editing or
post production, comes because of 1 bit sigma-delta converter by every
single bit by every requantisation process generates this noise again
and again.

Just one more indication of real life effect: Nyquist says _if_ the low
pass filter rings _then_ causes noise in this and this frequency . He
says not, that _it must ring_. With better technic you can avoid it. It
is up to You.

Quantisation distortion, the 1/2 LSB thing by _any_ bit rate _the last_
significant bit, by 16 bit or 24, 32 and so forth. Against this was
dithering born. Two pair different shues.

I would suggest first to clear and analyse these things and their impact
in the hearable range, before to scratch by -150dB. That is a misleading
trap of 1 bit sigma-delta troups. Why? Because by 1 bit the first bit is
the LSB at the same time! They _must_ use dithering to achieve a
bearable sound. They _must_ use very intensive noise shaping algorithms,
which leads to the 150dB S/N datas.

That is the ground why you mix the two things. Am I right?

Best regards
--
Johann Spischak

SDG, Spischak Digital GmbH
+49-911-965-7319
http://sdg-master.com

"William Sommerwerck" schrieb im Newsbeitrag
. ..
An example: Digital signal is aritmic, not systematic impulses. Two
or three contigous 1-s or 0-s should mean DC current for that time.
The overswinging in a transistor after the switching rings out. The
power supply can answer the next request. This was the ground
to develop the EFM technic [sic], by production of optical discs.

EFM is (eight-to-fourteen modulation) is a method to increase data
robustness. It has nothing to do with controlling the load on the
power
supply.

It has a lot to do with it. Can You explain how data looks if it is not
robust? What is the ground for it? I think You will land by my
sentences.

Best regards
--
Johann Spischak

SDG, Spischak Digital GmbH
+49-911-965-7319
http://sdg-master.com

Pooh Bear wrote:


Dirk Bruere at Neopax wrote:


If they chose to pay on "results", you are ****ed.

Let me put it this way...
We had one guy walk in, listen to some prototypes and immediately asked to be
put at the top of the list when they are sold. They were not even fully
equalised across the spectrum at the time, just hand tweaked on the bass/treble
knobs of the driving amp.


You mean Dobson's ultra-heavy bass boost ?

No.
These were without a sub or any significant bass!

They *will* sell - I have no doubt about that at all. Everyone who hears them is
really impressed. Plus, they look amazing.


I know, I designed them.

I very much doubt it.
What do you think I'm talking about?

Look where I am now !

Hanging round SED?

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org

"Chel van Gennip" schrieb im Newsbeitrag
...
On Thu, 19 Jan 2006 15:22:13 +0100, Johann Spischak wrote:

You are mixing two different things. Quantisation noise roots from
the
ringing of the low pass filter in the ADC, raises according

Not but you are mixing different things:
http://www.digitalradiotech.co.uk/sampling.htm
Because the digital sample values are discrete there will be a small
error between the actual analogue voltage and the encoded amplitude.
This error is called quantisation noise. Increasing the number of
bits
of the ADC decreases the voltage difference between adjacent levels
(increases the resolution) and therefore the quantisation noise is
reduced. The maximum value of the quantisation error is equal to
half
the voltage of the least significant bit.

Chel, You have obviously the false information source. But read it
again: it says noise to the quantisation error, which is not noise but
distortion. The noise will be shaped (transferred to the non hearable
range), the disortion however will be covered by additional noise
(dithered). Two different problems, fighted with two different weapons.

I have read that text on your suggested link.now and see, that the
person who wrote it made this mixing several times within one single
page. Obviously he/she has no idea, but edited this text from some
different prospects. Based on Your previous wise posts I expect from You
to recognise it. Please read it again.

Best regards
--
Johann Spischak

SDG, Spischak Digital GmbH
+49-911-965-7319
http://sdg-master.com

"Johann Spischak" wrote in message
...
"Chel van Gennip" schrieb im Newsbeitrag
...
On Thu, 19 Jan 2006 15:22:13 +0100, Johann Spischak wrote:

You are mixing two different things. Quantisation noise roots from the
ringing of the low pass filter in the ADC, raises according

Not but you are mixing different things:
http://www.digitalradiotech.co.uk/sampling.htm

Because the digital sample values are discrete there will be a small
error between the actual analogue voltage and the encoded amplitude.
This error is called quantisation noise. Increasing the number of bits
of the ADC decreases the voltage difference between adjacent levels
(increases the resolution) and therefore the quantisation noise is
reduced. The maximum value of the quantisation error is equal to half
the voltage of the least significant bit.

This is a fair account of the orthodox and accepted basics of digital audio.
However calling quantization noise a noise while conventional is not
perfectly accurate. Quantization error is related to the sample rate and the
signal being quantized, so it might be better to call it a distortion. It's
perfectly predictable from the signal and sample rate, and it is periodic
for periodic signals.

Chel, You have obviously the false information source. But read it again:
it says noise to the quantisation error, which is not noise but
distortion. The noise will be shaped (transferred to the non hearable
range), the disortion however will be covered by additional noise
(dithered). Two different problems, fighted with two different weapons.

Actually quantization error can be spectrally shaped, but one permissable
shape is flat response.

Dither does not cover noise, it randomizes or more specifically
pseudo-randomizes the quantization error.

Quantization error, being a form of distortion that is based on the sample
rate and the signal being digitized, is a coherent periodic signal when the
signal being digitized is coherent and periodic. Most music has periodic
components for perceptually significant periods of time. Therefore the
quantization error signal itself can be periodic, in-band and potentially
audible. When audible quantization error can be actually quite ugly-sounding
because it is usually not harmonically related to the signal that causes it.

The key think to remember is that dither does not "cover up" quantization
error. Dither randomizes quantization error and makes it less ugly to listen
to. Combined with spectral shaping, dither can cut quantization error's
audibility by *very* signficant amounts. Clearly audible quantization error
can be made quite palatable with the skillful application of dither and
spectral shaping.

In 16 bit systems, quantization error can be difficult to hear, even without
dither. Reasonble application of spectral shaping and dither can make
otherwise good 16 bit systems sonically indistinguishable from the
proverbial straight wire.

Arny Krueger wrote:

I took a look at the guy's web site and its full of self-serving BS.

http://www.grimmaudio.com/whitepaper...%20recipes.pdf

Let's start with:

"All the finest theory cannot predict how a human listener will experience
music as reproduced by a particular circuit."

Yecch!

The guy obviously has no idea about what audio theory can predict starting
with Bell Labs in the late 1920s and early 1930s through Zwicker and Fastl
in the 1990s. Actually, he probabaly does know this stuff perfectly well,
but puts his pontification suit on when he wants to sell his schtick to
unsuspecting rubes.

Eelco is so far ahead of you in audio work that in comparison you are a
turtle with a bicycle.

--
ha

"Arny Krueger" schrieb im Newsbeitrag
...

"Johann Spischak" wrote in message
...
"Chel van Gennip" schrieb im Newsbeitrag
...
On Thu, 19 Jan 2006 15:22:13 +0100, Johann Spischak wrote:

You are mixing two different things. Quantisation noise roots from
the
ringing of the low pass filter in the ADC, raises according

Not but you are mixing different things:
http://www.digitalradiotech.co.uk/sampling.htm

Because the digital sample values are discrete there will be a
small
error between the actual analogue voltage and the encoded
amplitude.
This error is called quantisation noise. Increasing the number of
bits
of the ADC decreases the voltage difference between adjacent
levels
(increases the resolution) and therefore the quantisation noise is
reduced. The maximum value of the quantisation error is equal to
half the voltage of the least significant bit.

This is a fair account of the orthodox and accepted basics of digital
audio. However calling quantization noise a noise while conventional
is not perfectly accurate. Quantization error is related to the sample
rate and the signal being quantized, so it might be better to call it
a distortion. It's perfectly predictable from the signal and sample
rate, and it is periodic for periodic signals.

Chel, You have obviously the false information source. But read it
again: it says noise to the quantisation error, which is not noise
but distortion. The noise will be shaped (transferred to the non
hearable range), the disortion however will be covered by additional
noise (dithered). Two different problems, fighted with two different
weapons.

Actually quantization error can be spectrally shaped, but one
permissable shape is flat response.

Dither does not cover noise, it randomizes or more specifically
pseudo-randomizes the quantization error.

Quantization error, being a form of distortion that is based on the
sample rate and the signal being digitized, is a coherent periodic
signal when the signal being digitized is coherent and periodic. Most
music has periodic components for perceptually significant periods of
time. Therefore the quantization error signal itself can be periodic,
in-band and potentially audible. When audible quantization error can
be actually quite ugly-sounding because it is usually not harmonically
related to the signal that causes it.

The key think to remember is that dither does not "cover up"
quantization error. Dither randomizes quantization error and makes it
less ugly to listen to. Combined with spectral shaping, dither can cut
quantization error's audibility by *very* signficant amounts. Clearly
audible quantization error can be made quite palatable with the
skillful application of dither and spectral shaping.

In 16 bit systems, quantization error can be difficult to hear, even
without dither. Reasonble application of spectral shaping and dither
can make otherwise good 16 bit systems sonically indistinguishable
from the proverbial straight wire.

Thank You Arny. There is only one point, where it still can be
misunderstood: You are thinking hear about "dynamic dithering", which
toggles only "when if". It will be used even up to 8 bits. (Brrrrrr) The
original dither toggles always, therefore it is an additional noise.
(fortunately not often used today)
If the ADC is not a 1bit sigma/delta one, You do not need to bother with
it.

Best regards
--
Johann Spischak

SDG, Spischak Digital GmbH
+49-911-965-7319
http://sdg-master.com

hank alrich wrote:

Eelco is so far ahead of you in audio work that in comparison you are a
turtle with a bicycle.

Ahead of Arnie? That can't be possible.

"hank alrich" wrote in message
. ..
Arny Krueger wrote:

I took a look at the guy's web site and its full of self-serving BS.

http://www.grimmaudio.com/whitepaper...%20recipes.pdf

Let's start with:

"All the finest theory cannot predict how a human listener will
experience
music as reproduced by a particular circuit."

Yecch!

The guy obviously has no idea about what audio theory can predict
starting
with Bell Labs in the late 1920s and early 1930s through Zwicker and
Fastl
in the 1990s. Actually, he probabaly does know this stuff perfectly well,
but puts his pontification suit on when he wants to sell his schtick to
unsuspecting rubes.

Eelco is so far ahead of you in audio work that in comparison you are a
turtle with a bicycle.

Certainly, I don't have a chance of catching up with his line of BS.

"Johann Spischak" wrote in message
...
"Arny Krueger" schrieb im Newsbeitrag
...

"Johann Spischak" wrote in message
...
"Chel van Gennip" schrieb im Newsbeitrag
...
On Thu, 19 Jan 2006 15:22:13 +0100, Johann Spischak wrote:

You are mixing two different things. Quantisation noise roots from the
ringing of the low pass filter in the ADC, raises according

Not but you are mixing different things:
http://www.digitalradiotech.co.uk/sampling.htm

Because the digital sample values are discrete there will be a small
error between the actual analogue voltage and the encoded amplitude.
This error is called quantisation noise. Increasing the number of
bits
of the ADC decreases the voltage difference between adjacent levels
(increases the resolution) and therefore the quantisation noise is
reduced. The maximum value of the quantisation error is equal to half
the voltage of the least significant bit.

This is a fair account of the orthodox and accepted basics of digital
audio. However calling quantization noise a noise while conventional is
not perfectly accurate. Quantization error is related to the sample rate
and the signal being quantized, so it might be better to call it a
distortion. It's perfectly predictable from the signal and sample rate,
and it is periodic for periodic signals.

Chel, You have obviously the false information source. But read it
again: it says noise to the quantisation error, which is not noise but
distortion. The noise will be shaped (transferred to the non hearable
range), the disortion however will be covered by additional noise
(dithered). Two different problems, fighted with two different weapons.

Actually quantization error can be spectrally shaped, but one permissable
shape is flat response.

Dither does not cover noise, it randomizes or more specifically
pseudo-randomizes the quantization error.

Quantization error, being a form of distortion that is based on the
sample rate and the signal being digitized, is a coherent periodic signal
when the signal being digitized is coherent and periodic. Most music has
periodic components for perceptually significant periods of time.
Therefore the quantization error signal itself can be periodic, in-band
and potentially audible. When audible quantization error can be actually
quite ugly-sounding because it is usually not harmonically related to the
signal that causes it.

The key think to remember is that dither does not "cover up" quantization
error. Dither randomizes quantization error and makes it less ugly to
listen to. Combined with spectral shaping, dither can cut quantization
error's audibility by *very* signficant amounts. Clearly audible
quantization error can be made quite palatable with the skillful
application of dither and spectral shaping.

In 16 bit systems, quantization error can be difficult to hear, even
without dither. Reasonble application of spectral shaping and dither can
make otherwise good 16 bit systems sonically indistinguishable from the
proverbial straight wire.

Thank You Arny. There is only one point, where it still can be
misunderstood: You are thinking hear about "dynamic dithering", which
toggles only "when if". It will be used even up to 8 bits. (Brrrrrr)

Not at all.

The original dither toggles always, therefore it is an additional noise.

That's what I'm talking about. However, that amount of noise is of
insufficent amplitude to mask quantization error. Dither is effective
because of how it interacts with the quantization process.

Johann Spischak wrote:

That is the ground why you mix the two things. Am I right?

Sorry Johann, I've been trying to follow you but as an electrical and
dsp engineer I can't figure out what you are trying to say.

Statements like, "Quantisation noise roots from the
ringing of the low pass filter in the ADC, raises according and
depending on the loudness." don't convey any meaning that I can decode.

I understand that there is probably a language barrier here, and
sympathize, but it's a bit too high for comprehension.


Bob
--

"Things should be described as simply as possible, but no simpler."

A. Einstein

"Arny Krueger" schrieb im Newsbeitrag
...

The original dither toggles always, therefore it is an additional
noise.

That's what I'm talking about. However, that amount of noise is of
insufficent amplitude to mask quantization error. Dither is effective
because of how it interacts with the quantization process.

The original intention was to cover the distortion with the noise. By
that specific one by the last bit has it worked, but this was only the
mouse. The elephant stays there, which is not possible to fight with
dither: The switching distortion. One of the good explanations with
clean graphs have we placed here, do not worry it is english:
http://sdg-master.com/lesestoff/DitherExplained.pdf

You will like it. These materials are on our site to help to find out
where took the development the false paths.

Best regards
--
Johann Spischak

SDG, Spischak Digital GmbH
+49-911-965-7319
http://sdg-master.com

Pooh Bear wrote:


Dirk Bruere at Neopax wrote:


Pooh Bear wrote:


Dirk Bruere at Neopax wrote:



If they chose to pay on "results", you are ****ed.

Let me put it this way...
We had one guy walk in, listen to some prototypes and immediately asked to be
put at the top of the list when they are sold. They were not even fully
equalised across the spectrum at the time, just hand tweaked on the bass/treble
knobs of the driving amp.


You mean Dobson's ultra-heavy bass boost ?

No.
These were without a sub or any significant bass!


They *will* sell - I have no doubt about that at all. Everyone who hears them is
really impressed. Plus, they look amazing.


I know, I designed them.

I very much doubt it.
What do you think I'm talking about?


Sounds like those 'column' speakers to me.

What do column speakers do?

--
Dirk

The Consensus:-
The political party for the new millenium
http://www.theconsensus.org

On Fri, 20 Jan 2006 02:46:33 +0100, "Johann Spischak"
wrote:

The original intention was to cover the distortion with the noise. By
that specific one by the last bit has it worked, but this was only the
mouse. The elephant stays there, which is not possible to fight with
dither: The switching distortion.

This is incorrect, however translated.

And the rest is boring.
JMO,

Chris Hornbeck

"Chel van Gennip" schrieb im Newsbeitrag
...
On Thu, 19 Jan 2006 17:03:53 +0100, Johann Spischak wrote:

Chel, You have obviously the false information source. But read it
again: it says noise to the quantisation error, which is not noise
but
distortion.

Johann, I think there is some misunderstanding in naming conventions.
Many
sources however use the name quantization noise for the errors
introduced
by the finite wordlength:
http://cnx.rice.edu/content/m11923/latest/
http://www.unizh.ch/phar/sleepcd/dem...hap3/digi5.htm The
effect
of the discrete steps in quantization, wether you call it noise or
error,
has a magnitude of about 1/2 LSB or about -150dB FS for a 24bit
digital
signal. I don't see that as a problem. It is a theoretical problem as
in
a practical 24 bit signal the lower 2,3 or 4 bits are pure noise.

The effect you described, imperfections of the low-pass filter, is a
completely different effect. Two imperfections can exist:
imperfections
below and above the filter frequency. Frequencies higher than the half
of
the sampling frequency passing though the filter to the quantisation
stage
will generate very nasty effects, resulting in errors/noise with
frequencies below the half of the sampling frequency. Imperfections in
the
filter at frequencies below the half of the sampling frequency are
much
less dramatic. Passing only 95% in a not to narrow band in the lower
range
wil not be a disaster, passing 5% anywhere in the high band is a real
disaster. So the filter must be well designed.

The separation line between quantisation distortion and quantisation
noise is there, where the origins of the two phenomenons are. About the
distortion part and the method against it you can find he
http://www.mtsu.edu/~dsmitche/rim420...20_Dither.html and
he
http://sdg-master.com/lesestoff/DitherExplained.pdf

Quantisation noise has three elements: The noise of the alias
frequencies, granulation noise and noise modulations. All of them are
depending from the level of the input signal. A louder signal will be
escorted by higher noise level in comparison to a week input signal.
This noise has no correlation with the signal, meanwhile by quantisation
distortion it is the case. This noise can be handled as a parallel noise
to the signal, can be separately transferred to a higher frequency
range, even beyond the Nyquist frequency.

The sense of a higher sampling rate is to be able to make less agressive
anti-aliasing filters, since they have also their own negativ influence
in sound quality. Here we could begin a fruitless discussion about FIR,
IIR, Brickwall or not and so on.

Both things are in the praxis the real disadvantages of the 1 bit
technic and its family: the sigma-delta pseudo multibit versions. An
modern R2R with a contemporary power supply is another world.

I have not studied the problem in depth, but I think PCM is superior
for
at least manupilating signals because any x-bit sample is a
representation
of an air pressure at a certain moment and sound is a fluctuation of
air
pressure over time. Just think about a normal situation: you have
three
sources A,B,C recorded with sufficient headroom. You want to sum these
sources into a final result: 2.67*A + 1.35*B + 0.77*C. A quite normal
operation. When you are using PCM signals, e.g. 24 bits samples stored
as
floats, the operation is a trivial processing per sample without
introducing significant errors in the process. If you try to do the
same
with DSD streams, the process is not simple anymore.

The main problem with the DSD method is, that it has no concret
relationship with 0. Therefore it is not possible to regulate, mix or
master it. It is possible to record it and play it. For production
purposes has Sony Oxford made a soundcard with 3x8 bit PCM intern. But
in their case I am not up to date, since I am not interested. Post
production will be made with different methods, like converting to 96/24
PCM and back (Universal) or to analog and back (Telarc). On the market
is no real DSD made SACD to find, except the demo discs played by AES
conventions.

Best regards
--
Johann Spischak

SDG, Spischak Digital GmbH
+49-911-965-7319
http://sdg-master.com

"Johann Spischak" wrote in message
...
"Arny Krueger" schrieb im Newsbeitrag
...

The original dither toggles always, therefore it is an additional
noise.

That's what I'm talking about. However, that amount of noise is of
insufficent amplitude to mask quantization error. Dither is effective
because of how it interacts with the quantization process.

The original intention was to cover the distortion with the noise.

You're the first person who has revealed that tidbit in my presence.

I wasn't there at the time, so I wouldn't know. Not that it matters one bit.

It seems to me that mechnical analog computers were being widely used at the
time and earlier by the US military as well (aiming weapons on ships for
example), so similar observations might have been made in the US at the same
time or earlier.

By that specific one by the last bit has it worked, but this was only the
mouse. The elephant stays there, which is not possible to fight with
dither: The switching distortion. One of the good explanations with clean
graphs have we placed here, do not worry it is english:
http://sdg-master.com/lesestoff/DitherExplained.pdf

Too bad about the nasty artifacts in Figure 14 ;-(

Other than that, its pretty much the usual story.

wrote:

The original intention was to cover the distortion with the noise. By
that specific one by the last bit has it worked, but this was only the
mouse. The elephant stays there, which is not possible to fight with
dither: The switching distortion.

No. Not at all. The original intention _was_ to linearize the system.
The mathematics of dither were worked out in the 1950s by the Bell Lab
folks.

And yes, dither DOES fight crossover distortion. That's the main benefit,
in fact.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Chel van Gennip wrote:
As the benefits of PCM, if you want to process a signal, are so obvious,
and as I never have heard of any solution for DSD that does not include an
intermediate PCM or analog format, I too am not so interested in any minor
developments there. It is a major DSD problem both on the production side
and on the consumer side.

DSD makes intermediate processing extremely difficult. As for me, I
would consider this a major advantage. I'd like to see most of the
signal processing stuff made totally impossible, even. Sometimes I
think it's all (even EQ) more trouble than it's worth.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

In article , (Scott Dorsey)
wrote:

Chel van Gennip wrote:
As the benefits of PCM, if you want to process a signal, are so obvious,
and as I never have heard of any solution for DSD that does not include an
intermediate PCM or analog format, I too am not so interested in any minor
developments there. It is a major DSD problem both on the production side
and on the consumer side.

DSD makes intermediate processing extremely difficult. As for me, I
would consider this a major advantage. I'd like to see most of the
signal processing stuff made totally impossible, even. Sometimes I
think it's all (even EQ) more trouble than it's worth.
--scott

I'll wager you're in the minority on this.

-Jay
--
x------- Jay Kadis ------- x---- Jay's Attic Studio ------x
x Lecturer, Audio Engineer x Dexter Records x
x CCRMA, Stanford University x http://www.offbeats.com/ x
x---------- http://ccrma.stanford.edu/~jay/ ------------x

Jay Kadis wrote:
In article , (Scott Dorsey)
wrote:

Chel van Gennip wrote:
As the benefits of PCM, if you want to process a signal, are so obvious,
and as I never have heard of any solution for DSD that does not include an
intermediate PCM or analog format, I too am not so interested in any minor
developments there. It is a major DSD problem both on the production side
and on the consumer side.

DSD makes intermediate processing extremely difficult. As for me, I
would consider this a major advantage. I'd like to see most of the
signal processing stuff made totally impossible, even. Sometimes I
think it's all (even EQ) more trouble than it's worth.

I'll wager you're in the minority on this.

Absolutely! Nobody ever wants to give up stuff that makes their life
easier. Sometimes their results are better when they do, though.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."

In article , (Scott Dorsey)
wrote:

Jay Kadis wrote:
In article , (Scott Dorsey)
wrote:

Chel van Gennip wrote:
As the benefits of PCM, if you want to process a signal, are so obvious,
and as I never have heard of any solution for DSD that does not include an
intermediate PCM or analog format, I too am not so interested in any minor
developments there. It is a major DSD problem both on the production side
and on the consumer side.

DSD makes intermediate processing extremely difficult. As for me, I
would consider this a major advantage. I'd like to see most of the
signal processing stuff made totally impossible, even. Sometimes I
think it's all (even EQ) more trouble than it's worth.

I'll wager you're in the minority on this.

Absolutely! Nobody ever wants to give up stuff that makes their life
easier. Sometimes their results are better when they do, though.
--scott

In an ideal world, we'd all make perfect recordings and none of this would be
necessary. Occasionally I have made recordings that don't need any "help". But
on a day-to-day level, some of the fun associated with doing audio involves how
much we can "fix" imperfect recordings within the limitations of our tools.

Still, for the purists, we can all marvel at how amazing recordings have been
made without any of the usual processing. It just doesn't happen often enough
to eliminate the need for the tools.

-Jay
--
x------- Jay Kadis ------- x---- Jay's Attic Studio ------x
x Lecturer, Audio Engineer x Dexter Records x
x CCRMA, Stanford University x http://www.offbeats.com/ x
x---------- http://ccrma.stanford.edu/~jay/ ------------x

In article ,
Chel van Gennip wrote:

On Fri, 20 Jan 2006 19:41:37 +0100, Jay Kadis wrote:

In an ideal world, we'd all make perfect recordings and none of this
would be necessary. Occasionally I have made recordings that don't need
any "help". But on a day-to-day level, some of the fun associated with
doing audio involves how much we can "fix" imperfect recordings within
the limitations of our tools.

Still, for the purists, we can all marvel at how amazing recordings have
been made without any of the usual processing. It just doesn't happen
often enough to eliminate the need for the tools.

There was a time when direct cut LP's were populair. I like to record live
performances as there is no substiture for the real adrenaline of a live
performance. I like to record with a safe headroom, as it hard to predict
how much emotion/volume there will be during a performance. So I don't
like a (DSD) recording system that does not allow for volume adjustment
afterwards.


http://www.superaudiocenter.com/Products.htm

-Jay

--
x------- Jay Kadis ------- x---- Jay's Attic Studio ------x
x Lecturer, Audio Engineer x Dexter Records x
x CCRMA, Stanford University x http://www.offbeats.com/ x
x---------- http://ccrma.stanford.edu/~jay/ ------------x