> My preference would be for 64K/24-bit for storage since we really can't do much
> better than 20-21 bits now and I think I've heard that thermal noise and such
> will limit the ability to ever reach true 24-bit quality in real-world
> ADCs/DACs. (On the other hand, true 16-bit quality was close to impossible a
so we'll soon get special temprature controlled DACs on your average
garden variety soundcards. Can't have precious bits go waste due to
temprture fuctuations now, can we? ;-)
> few decades ago...) Also, dynamic range of human hearing from threshold of
> audibility to threshold of pain is somewhere close to 24-bits. I wouldn't want
> to be burdened with 33% more data across the board just in case I wanted to
> reproduce both a flea and the big bang in the same performance!
You could probably rectify that with clever use of fp instead of integers -
but is it worth it?
>
> But in systems where bits are plentiful and 32-bit words are more convenient,
> then by all means.
>
--
Sander
+++ Out of cheese error +++
Reply by Sander Vesik●April 27, 20042004-04-27
Randy Yates <yates@ieee.org> wrote:
> Piergiorgio Sartor <piergiorgio.sartor@nexgo.REMOVE.THIS.de> writes:
> > [...]
> > independently from human perception capabilities
>
> Right. And we should start extending LCDs, CRTs, and other
> display devices to ultraviolet - you never know what the
> eye might do with those high frequencies.
A certain portion of hukmans can see low ultraviolet - it needs
a mechanical change that is in these cases done for unrelated to
ultraviolet seeing medical reasons - but CRTs and LCDs (esp LCDs!)
have a lot of other problems to be cured first before bothering with
that.
--
Sander
+++ Out of cheese error +++
Reply by Max Hauser●April 27, 20042004-04-27
"Randy Yates" in news:xxp1xmawqna.fsf@usrts005.corpusers.net:
> "Max Hauser" <maxREMOVE@THIStdl.com> writes:
>> [...]
>> ... "oversampling" and "noise shaping" are wide terms in connection
>> with A/D conversion, while "delta-sigma" (coined and popularized
>> by Inose et al., 1962 -- the alternative form "sigma-delta" appeared
>> later, about 1974) always meant systems with specifically _one-bit_
>> quantization.
>
> Are you referring to:
>
> "The Subscriber-Line Circuit and the Signaling and Tone System for
> an Experimental Time-Division Exchange Featuring Delta-Modulation
> Techniques," H. INOSE et al.
No, the following two papers:
Inose, Yasuda, and Murakami, "A Telemetering System by Code Modulation --
Delta-Sigma Modulation," _IRE Transactions on Space Electronics and
Telemetry_ vol. SET-8 pp. 204-209, September 1962. (Reprinted in N. S.
Jayant's IEEE-Press reprint volume on _Waveform Quantization and Coding,_
1976.)
Inose and Yasuda, "A Unity Bit Coding Method by Negative Feedback,"
_Proceedings of the IEEE_ vol. 51 pp. 1524-1535, November 1963.
It happens I just fetched those two references from a message I posted here
on dangers of "white" quantization noise models -- on comp.dsp, message
<51304@prls.UUCP>, 6 Aug 1991. The same posting also addressed the
following recent separate question
"Funky" in news:c6hlp9$ag9$1@newsg4.svr.pol.co.uk...
> Which publication did Shannon prove the Sampling Theorem in?
-- as it happens, message <51304@prls.UUCP> was an earlier screed against
bizarre practices to be found in modern engineering research, with positive
references, ending in reply to something else
> And, of course, there is no "Nyquist theorem" (except
> maybe on the Usenet) -- the honorific idiom is Nyquist rate,
> and the theorem is the sampling theorem (all of this was
> coined by Claude Shannon in "Communication in the
> Presence of Noise," _Proceedings of the IRE_ vol. 37
> no. 1 pp. 10-21, January 1949).
Two current queries answered with one 1991 posting here. How's that for
promptness?
Reply by Carlos Moreno●April 26, 20042004-04-26
Max Hauser wrote:
> "Carlos Moreno" in news:uSYic.24731$2V6.341047@wagner.videotron.net...
>
>>Wait... Are you pulling Bob's ear and telling him "bad Bob"
>>for quoting too much of your message for the little he wrote,
>>or are you answering his question saying that you wrote that
>>much because you had to exceed what you had quoted?
>>:-)
>>Carlos
>
> Exactly!
See, I knew I had gotten it all wrong! Thanks for correcting
my mistake! :-)
Carlos
--
Reply by ●April 26, 20042004-04-26
"Max Hauser" <maxREMOVE@THIStdl.com> writes:
> [...]
> Speaking of language, you probably know (but some recent authors appear not
> to) that "oversampling" and "noise shaping" are wide terms in connection
> with A/D conversion, while "delta-sigma" (coined and popularized by Inose et
> al., 1962 -- the alternative form "sigma-delta" appeared later, about 1974)
> always meant systems with specifically _one-bit_ quantization.
Are you referring to:
"The Subscriber-Line Circuit and the Signaling and Tone System for
an Experimental Time-Division Exchange Featuring Delta-Modulation
Techniques," H. INOSE et al.
?
--
Randy Yates
Sony Ericsson Mobile Communications
Research Triangle Park, NC, USA
randy.yates@sonyericsson.com, 919-472-1124
Reply by Max Hauser●April 26, 20042004-04-26
"Bob Cain" in news:c6ff6t0oe3@enews2.newsguy.com
> . . . [long original article snipped]
> Max, try as I might, I can't figure out what your point is
> with all this.
Sorry about previous response in limited time to a question I could not
immediately resolve as to seriousness, to an earlier dense post on my part.
I have appreciated many of Bob Cain's past, constructive posts here. (-:
Such as at least five recent and needed remonstrations about reposting a
long article merely to add a short reply. :-) But I especially resonated
with Bob's remark, 31 Jan 04, on technical authors who form a problem to
suit a solution rather than vice-versa (if I understood that point). I
have seen enough of such things in modern engineering research to believe
someone could write an interesting book on it, or at least an essay; however
this requires that they know many such cases, and people in that situation
usually have more constructive things to do. But to give a disinterested
example, a co-worker of mine some time ago was a regular co-editor of the
IEEE Transactions on Signal Processing (earlier Transactions on ASSP,
earlier still Transactions on Audio and Electroacoustics). This engineer
had substantive background in a major application of signal processing, and
complained once about a paper claiming important algorithm results, but
using, he said, wholly unrealistic assumptions about the application,
obvious to people with experience in it, but permitting sweeping and elegant
results. (I am omitting the application area here, it does not affect the
substance of the case.) Yet this paper was evidently sincere, and reflected
the work of someone of skill and standing whose focus had perhaps favored
publications per se over practice. When these situations arise, a gulf of
intuitions or assumptions exists and it can be be hard for someone in my
co-worker's position to explain the problem to someone in his author's
position, who moreover is eager to see the paper in print.
Back to Bob Cain's question. The first part of that long post of mine,
<108lirvdplak2fd@corp.supernews.com>, answered Randy Yates's query whether
papers addressing "oversampling analog-digital interfaces" that actually
entailed quantization could address such interfaces as essentially a problem
in filtering. (Please remember that "oversampling analog-digital
interfaces" is an umbrella term I'm using that does imply different
emphasis to different people, but I am using it for situations that have, or
would be reasonably entail if actually used, both a "time" interface and a
"value" interface, between continuous and discrete.) I referred to
published writing on this situation in general. In the next paragraph I
cited language ambiguities in that specialty, and the practice in some
research of partitioning facets of a block that are less separate in
practice. I then raised the point that analog-digital interfaces for
signals, as distinct from sample-by-sample data, are more common
commercially now than several years ago, illustrating with light anecdotes
drawn from life. In the next, short paragraph I supported Yates's point
that oversampling done to simplify analog filtering hardware is indeed a
filtering issue. (It is that historical mission that flavored the tone of
some DSP researchers' approach to later oversampling analog-digital
interfaces that did include quantization. For example, a body of somewhat
isolated literature developed that addressed such interfaces in the A/D
direction as digital filters having "analog input," their term. This
occupies section 9.12 in the organized bibliography within my 1991 JAES
survey paper, and also appears in the review of origins of oversampling
analog-digital interfaces in section 6.3 there.)
My penultimate paragraph addressed a language ambiguity that is relatively
recent (mostly post-1991) by pointing to the record evident in the
literature, that the coinage "delta-sigma" was made to distinguish systems
built around one-bit quantization, and this was the meaning of "delta-sigma"
intended by writers for most of the history of that term. It mentioned in
passing a theoretical advantage that contributed to special interest in
one-bit quantization in monolithic oversampling data-conversion hardware
(besides a basic simplicity, which had longer motivated exploration of using
one-bit quantizers). This paragraph was, I concede, dense. I will add here
as supplemental explanation that the people who coined the language
"delta-sigma" (Inose and Yasuda), and their predecessor (Cutler) who
popularized the more generic technique of multi-bit noise-shaping
quantization (where the baseband content tracks the original signal), were
not trying to build A/D or D/A converters in the modern sense, let alone
full conversion or reconstruction chains, but rather were doing telemetry or
speech or video "coding" in which they directly transmitted the high-rate
oversampled noise-shaped signal derived from an analog input, then
analog-lowpass-filtered it at the receiving end to recover an analog output.
(This was proposed as a hardware-efficient alternative in 1954-63 to methods
like FM and "full" PCM.) The mathematical basis of these techniques
attracted renewed interest for the later problem of making manufacturable
analog-digital interfaces, especially in the 1980s, as it became practical
to make real-time specialized digital filters in monolithic form
(inconceivable in the days of Cutler or Inose and Yasuda), while at the same
time, analog realities for conventional high-resolution A/D -D/A conversion
and sharp-cutoff filtering in monolithic form motivated searches for
alternatives.
(The ultimate paragraph in my posting was an offer to assist any authors
unaware that using "delta-sigma" with other than one-bit quantization
situations actively promotes another ambiguity formerly absent, but who seek
to rationalize this usage. That is because I have extraordinary experience
with such rationalizations.)
So there were multiple points, and if any of them is still unclear, Bob, I
may possibly be able to try further. I think that this subject naturally
resists brevity and promotes density. (-: But remember, I expect
responses to exceed the length of quoted text! :-) I have more on
stranger-than-fiction realities in engineering literature today, but have I
not said enough here in one post?
Max Hauser
Reply by glen herrmannsfeldt●April 26, 20042004-04-26
Arthur wrote:
(snip)
> Thank you for all the peoples participated in the dicussion..
> The conclusion seems to be : 96 KHz + 24 bits should be enough for all
> sorts of reasons. But there is a theory around stated that sampling
> higher will lower down the noise floor.. Any thoughts about that..
Well, increasing the number of bits per sample will do that, and
there are tradeoffs between sampling rate and bits.
In the early days of oversampling CD players, one reason to do
it was that simpler (cheaper) D/A converters could be used
with equivalent results. I believe that, as a storage
medium, increasing the bits per sample is more efficient.
When the CD first came out digital processing was still expensive.
Note that CD's only allow 99 tracks, as the track number is stored
in 8 bits BCD so that CD players don't have to do a binary to BCD
conversion!
-- glen
Reply by Arthur●April 26, 20042004-04-26
"Arthur" <arthurc99@yahoo.com> wrote in message news:<c65haq$1dgf$1@news.hgc.com.hk>...
> Hello all,
>
> Recently, there is a tendency to use high sampling rate 96 KHz, 192 KHz in
> the audio application. To my knowledge, about ~ 20 KHz, everything will be
> cut off from the loudspeaker, amplifier, etc.. So, why 96 KHz, 192 KHz
> instead of 48 KHz..
>
> Regards
> Arthur
Thank you for all the peoples participated in the dicussion..
The conclusion seems to be : 96 KHz + 24 bits should be enough for all
sorts of reasons. But there is a theory around stated that sampling
higher will lower down the noise floor.. Any thoughts about that..
Regards
Arthur
Reply by Mark Ovchain●April 26, 20042004-04-26
danlavry@mindspring.com (dan lavry) wrote in message news:<673b149b.0404231149.142e40ed@posting.google.com>...
> mark_ovchain@yahoo.com (Mark Ovchain) wrote in message news:<d61ab25b.0404221910.11b30855@posting.google.com>...
> > "Arthur" <arthurc99@yahoo.com> wrote in message news:<c65haq$1dgf$1@news.hgc.com.hk>...
> > > Hello all,
> > Well, I suspect that it helps the manufacturers sell more equipment.
> >
> > Think of it as a form of "trickle-down", why don't you?
>
> Think of it as "trickle-down"? Like - they are standing on the roof
> with the zipper open, trickling down on their customers? :-)
>
> Dan Lavry
Hush, now, I wasn't going to explain that part yet!
Reply by Max Hauser●April 25, 20042004-04-25
"Carlos Moreno" in news:uSYic.24731$2V6.341047@wagner.videotron.net...
> Wait... Are you pulling Bob's ear and telling him "bad Bob"
> for quoting too much of your message for the little he wrote,
> or are you answering his question saying that you wrote that
> much because you had to exceed what you had quoted?
> :-)
> Carlos