rgaetz@hydra.unm.edu (Robert Gaetz) (03/19/91)
Hello. Some months ago I posted a possible suggestion as to why
certain types of black magic such as spray applications, green
markers, and damping rings could make a sonic difference in what
should, ideally, be a perfect data retrieval system. I had proposed
that it seemed possible that noise on the digital signal could get
through the DAC and end up bulrring its output just enough to be
audible. My only reply was a flame wondering how I could think such
a dumb thing since digital circuits rely on saturation logic and
so this can't be the case.
Well, I suppose it's easier to simply write off all these
reports of improved performance than to actually look into the
problem, but luckily one company did look into it. Recently, the
JVC XL-Z1010TN cd player was reviewed by TAS and the result was
surprising. In short, they felt it was _the_best_ cd player they've
ever heard. It seems JVC had their engineers look into these effects
based on the assumption that there was somthing making a difference.
Essentially, they found that noise on the digital signal seemed to
effect sound quality, so they built a circuit that took a quick
peek at the incomming digital signal to determine whether it was
a high or low then generated a completely new signal which went to
the DAC.
Now, I'm not trying to say that everyone out there who comes up
with a miracle cure for an alleged problem should be taken
seriously, certainly anyone who would spend $349.95 for a digital
clock (Douglas Adams would have a field day with this one!!) to
improve their stereo should think seriously about what they're
about to do. I tend to think there's overwhelming reason to think
there's some other factor contributing to the sound quality of
digital audio. In addition to the reasons I mentioned before, the
issue of 20th generation copies of digital material sounding
different seems to be a clear sign that noise is a contributing
factor to sound quality. If you listen to a 20th generation copy
of an analog signal, you'd probly find that, among other things,
it is a lot noiser than the original, so surely the data on a 20th
generation digital signal will contain much more noise than the
original.
I may only be an undergrad EE, but I don't think it's fair to
suggest that a person has to be a Physicist to be any good at
pool.
Robert Gaetzed@mtxinu.COM (Ed Gould) (03/20/91)
> If you listen to a 20th generation copy of an analog signal, you'd > probly find that, among other things, it is a lot noiser than the > original, so surely the data on a 20th generation digital signal > will contain much more noise than the original. Only if it's copied as if it were analog data. Correct copying of digital information should regenerate and reform the signal at every stage. After all, the beauty of digital information is that one can *know* what it's supposed to be. If the copied digital signal is not as good as the original, then there's something wrong with the copy. There is a separate question of error correction. Errors do happen in digital recording and transmission, but there are techniques (involving redundant information) to correct such errors. Some of those techniques have an overhead of just a few percent, so they're not a significant factor in storage space or transmission time. Does anyone know what - if any - error correction techniques are employed in digital audio? Note that it is *extremely* unlikely for errors to appear within logic circuits. For all practical purposes, they only happen in recording/playback and transmission. -- Ed Gould mt Xinu, 2560 Ninth St., Berkeley, CA 94710 USA ed@mtxinu.COM +1 415 644 0146 "I'll fight them as a woman, not a lady. I'll fight them as an engineer."
bill@uunet.UU.NET (Bill Vermillion) (03/20/91)
In article <10340@uwm.edu> rgaetz@hydra.unm.edu (Robert Gaetz) writes: > In addition to the reasons I mentioned before, the >issue of 20th generation copies of digital material sounding >different seems to be a clear sign that noise is a contributing >factor to sound quality. If you listen to a 20th generation copy >of an analog signal, you'd probly find that, among other things, >it is a lot noiser than the original, so surely the data on a 20th >generation digital signal will contain much more noise than the >original. The "20th generation" caught my eye here. Recently there has been much discussion of the article in Mix magazine whereby several "important" audio engineers listened to 20th generation copies made on DATs. However - these 20th generation copies WERE 20TH GENERATION ANALOG copies. The amazing part (to me at least) is that there was so little apparent differences, because analog has always been additive in it's noise/distortion in successive copies. It shows that the analog circuitry has been very well refined. It wasn't that much noisier, and the distortion levels were way up from the original they were still siginificantly low. Lower than many first generation analog tape recordings I have seen. -- Bill Vermillion - UUCP: uunet!tarpit!bilver!bill : bill@bilver.UUCP
david@agora.rain.com (David Robinson) (03/21/91)
I've wondered about the issue of multi-generation copies of digital sources.
Prior speculation about the subject brought just one private message deriding
the notion that digital could be anything less than perfect, since we all
*know* that the error checking can't be off and all that predictable
sputtering....
Here's a thought experiment: If algorythms are all that we need,{ shall
we assume that a 1,000th generation of an original 128x oversampled DDD
tape will sound EXACTLY like the master tape?
Anyone here willing to bet the farm that they would? And if they don't...
and my listening to the Chesky test CD indicates that this is so after "only"
100 generations...then how can that be?
And how could JVC's new CD unit be "that good"?
And how do we handle "perfect sound forever"...that keeps on getting
perfecter...???
;-)
david@agora.rain.com
--
My opinions belong to me...and vice versa. They're not copyrighted;
third party thinkers should feel free to clone them at will.
david@agora.rain.com davidr@glacier.UUCP exspes@gdr.bath.ac.uk (P E Smee) (03/22/91)
In article <10340@uwm.edu> rgaetz@hydra.unm.edu (Robert Gaetz) writes: > Well, I suppose it's easier to simply write off all these > reports of improved performance than to actually look into the >problem, but luckily one company did look into it. ... >surprising. In short, they felt it was _the_best_ cd player they've >ever heard. It seems JVC had their engineers look into these effects >based on the assumption that there was somthing making a difference. >... they built a circuit that took a quick >peek at the incomming digital signal to determine whether it was >a high or low then generated a completely new signal which went to >the DAC. > >In addition to the reasons I mentioned before, the >issue of 20th generation copies of digital material sounding >different seems to be a clear sign that noise is a contributing >factor to sound quality. If you listen to a 20th generation copy >of an analog signal, you'd probly find that, among other things, >it is a lot noiser than the original, so surely the data on a 20th >generation digital signal will contain much more noise than the >original. No, this last assertion is untrue. (I've quoted your first part, which is a real effect, because I'm going to use it against you.) Much like the regeneration in your JVC, when you perform a digital copy of a digital tape, the equipment reads the signal from the original tape, 'peeks at it' to determine whether it should be a one or a zero (high or low), and then generates a completely new signal to put onto the copy. Analog copies, on the other hand, have to take what they're given (including any tape noise and degradation imposed by the circuitry it's already been through) and put that signal out to the copy tape which will, of course, add its own tape noise on top. If there is some effect which makes digital copies sound different than the originals (an effect which I've been unable to spot myself) then whatever it is, it's not the type of noise you seem to have in mind. Now, there IS also one weakness with your JVC story, as well. That is that ALL cd players perform the same task. They peek at the incoming data, which will come from the tape heads (on a DAT) or from the laser scanner (in a CD) in an incredibly sloppy form, which bears about as much resemblance to a digital square wave as a packet of peanuts does to the State Home for the Criminally Insane (apologies to Douglas Adams). They decide whether it's supposed to be a high or a low, and generate a clean digital signal to send to the DAC. It's entirely possible that JVC have found a more noise-immune way of doing this. Meridian (to pick another example) have managed to convince themselves that noise generated by the DAC and later analog stages can feed back into the digital part of the circuitry to cause problems, and have designed a (very nice) CD player by concentrating on minimizing that. Or, in summary, there are effects which need looked at, but I think you're only half-way to understanding them at the moment. Keep plugging away at it. -- Paul Smee, Computing Service, University of Bristol, Bristol BS8 1UD, UK P.Smee@bristol.ac.uk - ..!uunet!ukc!bsmail!p.smee - Tel +44 272 303132