[Russ Arcuri]

Quote:

Originally posted by morphsci First of all mutations are not trully random in their occurance. DNA is not simply a string of A's, C's, G's and T's. There is secondary, tertiary and quartenary structure which limits areas wher muation is likely to occur.

Yes, I know -- that's why I specifically said that the changes in the genetic code for the purposes of this analogy were at random spots. I wasn't talking about real-life genetic mutations or drift. I'm asking you to accept a conjecture -- that in transcribing a DNA sequence, some of the adenine-thymine and guanine-cytocine pairs got switched, in totally random locations on the string. If such errors really could happen, there is no way that you could have non-random results, such as 'greener leaves.' Quote:

Even , if we ignore the above facts, your analogy is only valid if green leaves is a simple, single gene Mendelian trait. If it is not then there may be many genes underlying the trait and its expression is not a simple linear function of the underlying genes.

I was aware of that as well, which is why I said, "a specific gene or genes on a specific chromosome or chromosomes must have quite specific changes." We've drifted too far into genetics now; I was simply trying to find an analogy you would appreciate. I'd love to talk genetics with someone who really knows something about it; it's been a long while since I did. But it's getting late and I'm getting tired. Quote:

Just like in the case of the reproduction of digital data there are underlying nonlinearities because of all of the complex interactions in the reproduction chain (such as power supplies interacting with DAC's interacting with the digital bitstream, etc.). Those nonlinearities in gene expression and development are the stuff of evolution, in the case of digital reproduction they are variation among copies.

There are more analogs too. A single bit error in the position of the least significant bit of a given 16 bit sample would probably not be heard, even if it happened to thousands of samples in a row. This is analagous to a transcription error on a DNA strand in one of the "junk DNA" parts of the string, as long as you believe that there are areas of non-useful "junk" DNA in the first place. An error in the MOST significant bit of a given 16 bit sample would be audible if it was duplicated in the same way -- the most signficant bit being wrong in several thousand samples in a row would be VERY audible. The analogous situation in genetics would be a transcription error on an 'important' gene, that we have a clear understanding of the functions of.

Even so, none of what you've said can explain how a simple D-D copy, even with multiple random errors and interactions with the power supply, etc, could cause the resulting music to simply sound "brighter," or "darker," or whatever. Random noise inserted into the bitstream will have a distinctly non-musical effect, rather than the effects described above. Quote:

I cannot reread that particular book since I do not have it any longer. [...] However, the idea that people provide proof of their claims is not part of the scientific culture, that is the popular science culture.

But when the claims are wild and not supported by the current knowledge, nobody is going to bother with checking them unless the claimant checks it himself. The chapter in the aforementioned book titled "The Dragon in My Garage" is what I'm getting at here. If someone claims there's a real dragon in their garage, it is up to the person making the wild claim to provide extraordinary proof. Simply saying, "I heard him in there, and I saw his fiery breath" is nowhere near the kind of proof that's going to get a scientist interested in the contents of the garage. Sagan (obviously) does a much more thorough job of explaining this. Quote:

I never claimed that uncritical acceptance of every claim is how I proceed in my work.

I didn't really think you did, but you seemed to be promoting the idea for a while there. You remember, "a scientist must be open to all possibilities, etc." Quote:

What I did caution against is an inflexibilty in the acceptance of current theory in any field. Science does not progress in a smooth linear fashion, but rather in fits and bounds by those willing to go beyond the present theory in both thought and deed. Unrelenting acceptance of dogma is the realm of religion not science.

I agree with basically everything you said in the previous quote. And for now, I am going to bed. I'm tired and have spent way too much time here tonight. Glad to see this didn't degenerate into a flame war; it was looking a little bleak at a couple points. Cheers.

 

[Xevion]

Okay, I skipped reading most of this thread because I need to go to bed, so what I have to say may have already been said.

Ooh, I rhymed

.

An audio CD player works a lot differently then a data CD reader because the audio CD reader must attempt to read the data in a linear fashion. It can't go over the same thing twice. Whereas a CD-ROM on a computer is spinning a hell of a lot faster and allows for random acess. This lets the drive have far superior error correction, and as such is much less susceptible to errors.

Now, when you burn a CD-R, the higher the speed, the less perfect the burn, although it is obviously good enough to be exactly the same to a data drive in most cases, to a CD player, it might not be able to read the CD-R as well, because the laser might have not burned through the reflective material as well, so occasionally (Many times a second I would imagine) the audio CD player thinks a 0 is a 1 and vice versa, and it just has to libve with it because it can't go back and recheck it, it has to continue plodding along and make any error corrections on the fly, which could be wrong themselves.

 

[MacDEF]

Quote:

Originally posted by Russ Arcuri Glad to see this didn't degenerate into a flame war

It didn't?

 

[Russ Arcuri]

Quote:

Originally posted by MacDEF It didn't?

Ha! That was pretty mild, actually. I've seen far, far worse before, and my disagreement with morphsci somehow managed to retain real content throughout. I don't know if you read this whole thing, but the Margio Dots guy I referred to above got crushed by several people at once, and that one was mostly humorous, if you can believe it.

In any case, I really appreciate the friendly environment here; I would have dropped out of the discussion entirely if the content disappeared and it degenerated into a true flame fest. I've seen some pretty nasty stuff on the rec.audio groups. Not too long ago I actually told some guy over on the TiVo forums that he was either too stupid to own a TiVo or a troll. Of course, in that case everyone else agreed with me. And even that is mild compared to what you'll find at any time of day or night on comp.sys.mac.advocacy or any of the politically-oriented discussion groups.

Looking back on this thread again, morphsci and I both got some digs in, but they don't hold a candle to most of the crap you see on USENET nowadays.

 

[coolvij]

Russ - does your wife know that you are braggin about your old flames on-line?

 

[Russ Arcuri]

Quote:

Originally posted by coolvij Russ - does your wife know that you are braggin about your old flames on-line?

LOL, coolvij. Actually, she thinks I was nuts for staying up late last night to continue the debate with morphsci; she couldn't care less about whether CD-Rs can sound brighter than the source, or whether some guy I've never met thinks Sagan's book is worth reading.

She wasn't even interested in the specifics of our disagreement on experimental testing -- and she's a psychologist, who's had plenty of coursework in experimental statistics, and actually applies this stuff in her work.

I have a hard time walking away from a good debate, actually. What sucks is when it degenerates into a flame war as it inevitably does on the USENET groups I referred to above. I've learned to simply stay away from those groups, because there couldn't possibly be a bigger waste of time than a back-and-forth exchange of insults with a stranger. I've probably participated in (maybe) 4 or 5 of them since 1992, and have always dropped out soon after the name-calling began. Except for the guy with the TiVo -- he was clearly a troll.

 

[Dusty Chalk]

Quote:

Originally posted by Russ Arcuri Dusty, I never said you were... but you seemed to be stepping in to support the original claim that I was taking issue with. I guess we were simply arguing different points. While I think the audibility of jitter is questionable, I never said it was impossible. I was specifically addressing whether burned CD-Rs could sound brighter due to copying.

Well, first I was arguing inaccuracy, I.E. -- sounding different. But now that I think about it, how can it sound anything but brighter? If you pervert a sinewave in the aforementioned way (with jitter), you introduce harmonics, higher-order harmonics than the fundamental. So, going purely on sinewaves, it would therefore be brighter.

It shouldn't do near as much damage on a pure square wave...except that it wouldn't necessarily be purely symmetrical anymore, which means adding even-order harmonics...again, resulting in a (admittedly subtly) brighter sound.

I think it would make it sound brighter, going purely on an extrapolation of these two examples, but I would have to think it through.

 

[Russ Arcuri]

Quote:

Originally posted by DustyChalk Well, first I was arguing inaccuracy, I.E. -- sounding different. But now that I think about it, how can it sound anything but brighter? If you pervert a sinewave in the aforementioned way (with jitter), you introduce harmonics, higher-order harmonics than the fundamental.

Most of what I was addressing previously was not jitter -- it was whether random errors (ones changed to zeros, or vice-versa) in large enough amounts could cause the sound of the resulting music to be brighter. I don't believe it can, as previously and exhaustively explained -- random corruption of the data cannot result in non-random effects in the resulting music. "Brighter" is certainly a non-random effect -- some people pay for EQ just to make the music brighter.

Okay, now let's discuss jitter a bit. You're not actually perverting a sine wave with jitter -- the diagram in that first article is easily misinterpreted. Jitter is a property of the digital bitstream -- not of sound, per se. This is why jitter can easily be corrected, simply by re-writing the bitstream to new media in a very precise manner. Get what I'm saying here?

At this point I had typed up a rather long, rambling essay about how varying amounts of jitter might affect the sound and at what levels it might be audible. Suffice it to say that I could not think of any situation in which the music would simply be brightened (or, indeed, in which the EQ could be changed in a non-random manner).

The problem is that I am very tired (a result of staying up so late last night) and my essay was quite literally a mess. Plus, I hadn't said anything in it that you couldn't get from already-established sources. Even the articles Tim D posted links to, if I am remembering them correctly, did not claim that the EQ of a given recording could change due to jitter. So I deleted the essay (out of consideration for anyone bored enough to read it) and suggest that people read up on jitter from other sources. If anyone finds something with a cogent explanation for how jitter could result in changes in EQ, post a link here. I am heading off to bed soon.

 

[Dusty Chalk]

Quote:

Originally posted by Russ Arcuri ...You're not actually perverting a sine wave with jitter -- the diagram in that first article is easily misinterpreted. Jitter is a property of the digital bitstream -- not of sound, per se. This is why jitter can easily be corrected, simply by re-writing the bitstream to new media in a very precise manner...

That's just wrong. Jitter -- inasmuch as it matters -- is a property of the A/D and D/A processes. Since we are, for the most part, talking about CD playback, I'll leave out the A/D portion. The only reason doing anything to it (the datastream) in the digital realm matters is because it will ultimately affect the final D/A process. So I disagree with that first statement. That is exactly what you're doing. If you don't reconstruct the sampled waveform at exactly the same instances of time as they were sampled (relatively speaking, of course), you are going to pervert that waveform. That is jitter in a nutshell.

However, in deference to your sleep (and mine), I will take your advice and read all the attached articles in detail. Might take a while, though, my printer cartridges ran out...

 

[pigmode]

My CD images made with EAC and Feurio (4x) are very good. When I used EAC to extract .wav files and burned with EZ CD, the results were brighter than the source CDs.

 

[aos]

>If you don't reconstruct the sampled waveform at exactly the same instances of time as they were sampled (relatively speaking, of course), you are going to pervert that waveform.

That is true and should be quite obvious to anyone who knows how digital to analog conversion works. There is no question that jitter will be reflected in the reconstructed signal, there isn't any black magic or questionable science here. It's visible from the formulas or from the pictures. I think the part of reluctance of people to acknowledge existence of some things is that it is very easy to slip into dogmas (axioms, whatever), it happens to me all the time. Even people who should know better will fall into the trap of single-track thinking on some issues, probably because life is too short and knowledge is too vast to acquire and keep through the whole time. We keep thinking that digital is only 1's and 0's and as long as their values are preserved, everything should be working fine. But in some cases we are trying to equal the processes that are fundamentally different, just because both have some roots in the digital domain.

Say, someone might say how could two CDs sound different if they contain exactly the same data? After all, we deal with say stored programs and documents every day and we know if there were any errors, our programs would not run correctly, our documents would look different than they did and so on. We recognize for the possibility of errors on the storage and are aware that in normal circumstances all of those errors will be corrected. And I mean ALL. Not a single bit of error after recovery is allowed when computer programs are involved. And we know it doesn't happen on media in good condition. Stuff wouldn't work.

But storing plain text of an email in digital format is not the same as storing music in digital format. One is nothing more than conversion from one number base (say 26 letters + punctuation) to the binary number base. You convert it and write down the new values somewhere on some media. Storing and retrieving it is a complicated process involving very advanced science (lasers, precise step motors, integrated electronics, fiberoptics, the whole physics of making a chip, capacitors, dam construction techniques (say hydro power) or battery chemical formulation, lots of mathematics of every possible form and shape - yes, I mean a good portion of the complete knowledge of human species) - but this process is very transparent to end user. And the actual conversion to fully defined and complete information can be done by elementary school students. You feed it symbols (from one codebase) and you get new symbols (binary). And then reverse. Their size, shape and spacing don't matter as far as information is concerned - a word is a word is a word. In a sense, an email is a very simple thing, a collection of symbols, and that's all. The words mean different things to me and you but every sane person will agree that an email they saw yesterday is the same they're looking at today, pulled from the hard drive where it was stored overnight.

But music is a complex thing. It is not a crude symbolic representation of the state of our brain, that people invented in order to communicate. It is the cause of the state, not the state itself and not its associated symbol. Music is a change of air pressure. And while the effects of words and what they invoke in our brains is probably of same complexity as the effect of sounds, words and music are entirely different beasts. It is completely different process that adds another dimension of complexity.

In this particular case of jitter, maybe a good analogy is tv signal. Imagine two LCD screens connected each to its own computer. LCD has a finite and obviously discrete resolution (1024x768 for example), and is in this example being fed a digital signal containing picture. Pixels are spaced in equal distances and they usually are three separate elements (red, green, blue). Let's imagine the intensity of each of the sub-pixels can be 0-255 and that the actual intensity is the same for both monitors (compare to reference voltage for DAC chips). Picture on the PC is in digital format and is the same on both machines, no doubt. Digital transmission is perfect or with correctable errors and both LCD monitors get the same pixel information. The picture should be the same. But, now imagine one of the LCD screens does not have the uniform spacing of pixels. Pixels 200 and 201 are not 500 but 600 micrometers apart, 201 and 202 are 610, 270 and 271 are 456 apart, and so on, in whatever pattern you like. Would you say both pictures are still the same?

You probably would if the variations in distances are smaller than a certain threshold. Or if the pattern is non-periodic. Or whatever. Better still, imagine that the distance is changing with time. Say they are compressing and expaning at a slow rate of few Hz.

In audio, whether jitter is audible, and if so, in what form, is a different discussion than just that of its existance. Effects of filtering, intermodulation distortion in the amp etc. add to the complexity and also make it unlikely that the effects of jitter would be the same on two different systems (or even on the model of the same manufacturer given component tolerances). However, lots of people claim to be able to hear the differences. I always believe that if there is actually a difference, then it can certainly be described by science, we only didn't try hard enough, or the poeple who could do it just couldn't be bothered.

It is likely possible to analytically describe the generated waveform once jitter is introduced although that kind of mathematics would likely be too much for any living person, even for the simplest formulas for jitter. But what calculations people can't handle, computers can. A simulation could be done. Could be a very nice masters/PhD for someone.

 

[Tim D]

Computer data transfer is not bit perfect either...there are just SOOOO many layers of error correction. Hardware level, OS/Device driver level, application level, etc.

The fundamental premise for any existing error correction is error detection, than either a correction algorithm(but you can't apply a bandage to a missing limb), or information resending with transmission acknowledgment. Since audio is a real-time application, there are some obvious constraints. Another reality besides being real-time is that it is NOT mission critical. Which basically means the error correction can be lossy.

Anyone who thinks digital is perfect because their computer applications are perfect do not notice some obvious facts. The first being that your computer doesn't run perfectly which is why there is something called "backup" or "reboot". And that there are layers upon layers of hardware/software abstractions where there is usually error correction (sometimes to the point of redudancy) in each and every layer.

If you really think digital music is perfect, you should learn about the differences between UDP and TCP Internet protocols. UDP would be more akin to CD playback. UDP is an internet transmission protocol, it does not guarantee reliable transfer, yet it is still a digital transmission. Why use UDP at all than if TCP is more robust? Because UDP is often use in streaming media where it is a real-time non-mission critical application where you actually gain performance in *most* instances because of less overhead. These are probably the EXACT fundamental reasons why CD streaming transfers are done in a UDP fashion as opposed to TCP.

 

[gaineso]

After all of this discussion, I still am not convinced, or about to be, that straight D to D copying introduces errors.

Copying from my H-K FL8300 CDP digital out to my Phillips single tray CD-R/RW digital in. Or my HP N5270 DVD tray to my Backpack Data CD-R/RW, using Adaptec CD Copier. Or my SO's HP XL844 internal CD tray to internal CD-R/RW.

There isn't any conversion going on, just transfering a data stream. Witnout any conversion AD or DA I just don't see how ANY error could come into the picture.

Of course, the audio rig is copying at 1X. I use the computers at 2X primarily because if I copy at 4X they write to the HD first and then to the CD-R so it actually takes just as long. Multiple copies save time writing to the HD, but singles don't.

Please tell me why straight D to D, as described, makes any errors.

 

[Russ Arcuri]

A few more comments. First, DustyChalk wrote: Quote:

That's just wrong. Jitter -- inasmuch as it matters -- is a property of the A/D and D/A processes.

This statement is simply not true. Jitter is a property of the digital bitstream. A CD in an of itself has no D/A conversion -- it's simply a set of data. That data on the CD can be jittery -- bits can be spaced in time on the disc incorrectly -- hence Sony's efforts described previously to come up with a digital mastering process which lays down less jittery CDs. I only made this point in the first place because the graph shown in that first web reference can be easily misinterpreted. The distorted square wave shown on that diagram is an attempt to demonstrate distortions not in the actual sound, but in the timing of the bits being fed to the DAC.

Now, I do agree that this only has a real-world effect once the data undergoes D/A conversion. But when we're making a D-D copy (remember the original assertion), jitter can be increased or decreased -- but in no way that I can imagine thus far, will the resultant data result in brighter sounding music when fed through the DAC.

Second comment: The essay I alluded to above, but deleted because it was a rambling mess, contained many of the same musings that aos's essay above contains. I described what jitter was in a real-world sense, and made various attempts to come up with a logical explanation for how jitter could result in what sounds simply like a brighter recording of the original. I could come up with nothing that made any kind of sense. Again, since jitter is not consistent -- one bit can follow another either too close behind or too far behind -- the effects, while difficult to determine, will be random. You can pervert a sine wave (representative of a single frequency tone, in the simplest case) in either direction; the large-scale effect of random perturbations is going to be a muddying of the sound, not an overall brightening. And when the jitter is SO bad that it actually causes significant data errors (as opposed to timing errors), the results, as I have pointed out numerous times already, aren't going to resemble anything musical in the first place.

Let's use aos's LCD screen analogy. Is a random perturbation of the spacing of the individual pixels going to result in a screen that looks normal, except that it's simply brighter? This would require that all pixels smaller than the standard size represent dark colors, and all pixels larger than the standard size represent brighter colors. But there's nothing inherent in the 'screen jitter' which would cause this to be so. i.e. We'd fully expect that some of the shortened pixels be bright colors and some of them be dark. The analogy to music is apt -- we'd expect the 'screen jitter' to muddy the picture somewhat, depending on the severity, just as I would expect jitter in a digital audio bitstream to muddy the sound if it were severe enough to be heard at all. Quote:

But music is a complex thing. It is not a crude symbolic representation of the state of our brain, that people invented in order to communicate. It is the cause of the state, not the state itself and not its associated symbol. Music is a change of air pressure. And while the effects of words and what they invoke in our brains is probably of same complexity as the effect of sounds, words and music are entirely different beasts. It is completely different process that adds another dimension of complexity.

I don't see anything here that I necessarily disagree with. However, I do disagree with your implicit assumption that digital data representing text and digital data representing music are somehow fundamentally different. They're not. It's only after D/A conversion that the data becomes "a change of air pressure," and even then, the effects of jitter are ill-defined. Quote:

If you really think digital music is perfect, you should learn about the differences between UDP and TCP Internet protocols.

Who is "you" referred to in red? I'm still assuming it's me. If it is, please explain to me where you got the impression that I believe digital music is "perfect." I'm considering dropping out of this time-sucking discussion entirely -- I'm getting pretty tired of correcting people who insist on arguing against points I haven't made. I say "I believe conjecture A," and someone says, "but how can you possibly believe conjecture B?"

Besides, I know all about the differences between TCP and UDP; I've been a systems/network administrator, and (currently) e-mail administrator for about a decade now. The TCP/UDP analogy can be applied to digital audio, but since you neglected to explain how lost data (UDP packets which never make the destination, in this analogy) would result in brighter sounding music. It seems the only person actually interested in addressing that question (other than me) is DustyChalk. Everyone else seems interested only in striving to get me to accept that jitter can make music sound different -- something I've not only admitted is possible, but spent paragraphs explaining on this thread. I will be happy to continue discussing the question at hand -- but I will not be pigeonholed into defending assertions I never made in the first place. If there's any question whatsoever about what I've been discussing since the beginning of my involvement in this thread, please go back and reread it -- all of it. Thanks.

 

[Joe Bloggs]

I haven't read much of this thread, to tell the truth, and someone might have addressed this before, but in that case I'd just say it again--

CD-DA (audio CDs) are a different data format from data CDs--it has a less secure error correction scheme.

If you copy straight from one drive to another on the fly you're likely to introduce errors in the copy--it seems that ripping software has looser standards of accuracy than when you are reading data CDs. After all, if there are errors in an audio CD you can still hear it, but if there are errors in a data file it probably becomes corrupted (esp. if it's an executable)

My guess is that on the fly 2-drive copying is probably done in burst mode only--when you rip a CD in secure mode on EAC and it detects an error it would have to stop and re-read that section many times to determine what bits were really intended to be stored there. If this happens during a disc to disc copy the copying would stall and unless you have a burn-proof drive you'd get a coaster...

My experience with EAC is that there are very few discs that I have where the built-in C1 and C2 error correction of the CD is enough for all the bit rot caused by poor disc conditions and read errors to be corrected on the fly from the incoming bitstream, without having to re-read at least one or two sections. Where EAC does this to get the correct bits in a secure rip, a burst mode rip-and-copy would just copy the errors.

Convinced?