23 Sep
2007
23 Sep
'07
9:06 p.m.
On Tue, 22 Apr 2003 19:54:09 -0500
"Dustin Barlow" <duslow at hotmail.com> wrote:
<snip>
So..? Most PCM converters utilize a 1-bit stream also. Why not
utilize all the tools available for the task at hand?
As for processing, you can look at a PCM representation of a waveform
to ease the processing load and then just apply the changes to the
orignal DSD stream without ever having to process in the 1-bit domain
directly (which is way more processor intensive since you have to look
at a huge chunk of the stream in order to extract the amplitude data
that is available in each multi-bit sample).
IMHO, though, the hippest alternative at present is to process a DSD
stream in the analog domain and re-record it to DSD. This results in
a very "analog" sound. These days you can get analog gear with a
respectable dynamic range for a song (Mackie Onyx anyone?). When you
can get a 130 dB S/N ratio in the analog domain you really don't lose
too much converting back and forth from 1-bit domain. It's freakin
sweet!
If you haven't tried recording 1-bit. Do yourself a favor and demo
one of the new Korg recorders. It really is really good, no kidding.
~Maitland
I read an interesting article on Direct Stream
Digital (DSD) / Pulse Density
Modulation (PDM) entitled "A Better Mousetrap" by Brian Smithers in the May
2003 issue of Electronic Musician. Since, Brian did a good job explaining
PDM/DSD in quasi-layman terms, I'll just quote snippets from his article to
set the stage for my questions.
<snip>
DSD/PDM appears to be a superiour technique for
recording and playing audio
material.
Having been around digital audio and digital signal processing for over 10
years, I am still far from convinced.
Granted, this technology may never catch on
because of all the
hardware and software changes that would be required to mirror what a
typical PCM based DAW currently does. But, if DSD/PDM does catch on, and
DAWs start being produced, how will this effect current audio DSP
techniques?
I have not looked into the maths behind algorithm development in DSD/PDM,
but I doubt it is anywhere near as easy as with PCM.
The article mentions a program called Pyramix
(Windows) which features DSD
support. However, for Pyramix to do EQ, dynamics, reverb processing, and to
display waveforms and vu levels, it converts DSD to a "high quality" PCM
format.
That should tell you something :-). 
23 Sep
23 Sep
9:28 p.m.
First off, why are you reprising a 4 year old thread if you don't have
anything new to add?
Maitland Vaughan-Turner wrote:
So..? Most PCM converters utilize a 1-bit stream
also. Why not
utilize all the tools available for the task at hand?
Yes, but you need to analyze what you are doing or the next thing
you know you are paying mega dollars for triple gold plated single
direction, rare earth metal interconnect cables and other such
snake oil.
As for processing, you can look at a PCM
representation of a waveform
to ease the processing load and then just apply the changes to the
orignal DSD stream without ever having to process in the 1-bit domain
directly (which is way more processor intensive since you have to look
at a huge chunk of the stream in order to extract the amplitude data
that is available in each multi-bit sample).
Problem : the conversion from DSD and PCM is lossy [0], hence doing
DSD -> PCM -> DSD -> PCM is a bad idea.
IMHO, though, the hippest alternative at present is to
process a DSD
stream in the analog domain and re-record it to DSD.
Two problems:
- A-to-D converters are limited to about 20 bits of SNR due
to things like silicon junction noise and those noisy
electron thingys.
- All commonly used audio A-to-D converters are 1-bit so every
time you go to analogue and then into a digital effects box
you are doing another DSD to PCM conversion.
See reference [0] below.
This results in a very "analog" sound.
So, you have a way of objectively measuring how "analog" something
sounds? I'd be interested in hearing your methodology.
These days you can get analog gear with a
respectable dynamic range for a song (Mackie Onyx anyone?). When you
can get a 130 dB S/N ratio in the analog domain you really don't lose
too much converting back and forth from 1-bit domain.
Where is you analogue to digital converter which also has 130dB SNR?
Erik
[0] Why 1-Bit Sigma-Delta Conversion is Unsuitable for High-Quality
Applications,
Stanley P. Lipshitz and John Vanderkooy
http://sjeng.org/ftp/SACD.pdf
--
-----------------------------------------------------------------
Erik de Castro Lopo
-----------------------------------------------------------------
Fundamentalist : Someone who is colour blind and yet wants everyone
else to see the world with the same lack of colour.
9:35 p.m.
Erik de Castro Lopo wrote:
[0] Why 1-Bit Sigma-Delta Conversion is Unsuitable for
High-Quality
Applications,
Stanley P. Lipshitz and John Vanderkooy
http://sjeng.org/ftp/SACD.pdf
See also :
http://en.wikipedia.org/wiki/Direct_Stream_Digital
which links to a number of other papers.
Erik
--
-----------------------------------------------------------------
Erik de Castro Lopo
-----------------------------------------------------------------
"Windows and MacOS are products, contrived by engineers in the service
of specific companies. Unix, by contrast, is not so much a product as
it is a painstakingly compiled oral history of the hacker subculture."
-- N Stephenson
24 Sep
24 Sep
6:19 p.m.
Erik de Castro Lopo wrote:
[0] Why 1-Bit Sigma-Delta Conversion is Unsuitable for
High-Quality
I'm glad that I built good multi-bit DAC-box around very nice BB
PCM-63P. It's still among the best sounding converters. Unfortunately
that's not available any more. However, there are still some high
quality real multibit converters around.
When it comes to delta-sigma converters I'm mostly concerned about the
transient and some linearity properties of those. And in general the
commonly used switched-capacitor filters. One big reason for these
converters to exist I can see is the low cost and easy implementation of
the associated analog filters. Thus cheap (or cheaply made) devices are
likely to sound better.
In some cases, direct DSP processing of the DSD stream would be feasible.
- Jussi
7:35 p.m.
Jussi Laako wrote:
In some cases, direct DSP processing of the DSD stream
would be
feasible.
Really?
Current CPUs are designed to work on integer or float samples. They
are not designed to work on single bits. Attempting to do simgle bit
processing on regular CPUs will be slow.
That means the only sensible place to do DSD processing efficiently
is in silicon; either FPGAs or ASICs.
Erik
--
-----------------------------------------------------------------
Erik de Castro Lopo
-----------------------------------------------------------------
"The worst thing about productivity tools is that there's
always something in them that makes you less productive"
-- Chris Alfred Dec 15, 1997
6217
days inactive
6218
days old
linux-audio-dev@lists.linuxaudio.org
4 comments
3 participants
participants (3)
-
Erik de Castro Lopo -
Jussi Laako -
Maitland Vaughan-Turner