Luke's "are identical CDs identical?" thread

From Edison cylinders to pre-amps to ProTools: talk about it here.
User avatar
lukpac
Top Dog and Sellout
Posts: 4591
Joined: Wed Apr 02, 2003 11:51 pm
Location: Madison, WI
Contact:

Luke's "are identical CDs identical?" thread

Postby lukpac » Sat Mar 18, 2006 12:04 pm

I know I've touched on a lot of this in the past, but it keeps coming up, which really pisses me off. I made the mistake of stumbling over to SH.tv recently to check out a Beatles thread, and well, I ended up catching a lot of other stuff too. Stuff that just made me angry.

Before I get into the messy bits (ha - bits!), though, a mission statement of sorts - it isn't my goal (at least at this point) to claim that differences can't/don't exist. I've never heard differences between identical sources, but at least one friend of mine who isn't a quack has (or at least thinks he has!). Perhaps such a study can one day be properly conducted, but honestly at this time I don't care to debate the issue.

What I do care about, though, is the notion that differences *have* to exist, and if you don't hear them you're either deaf or have "inferior" equipment. My feeling is, and has always been, that if differences can exist, it's entirely based on your playback system. That is to say, two discs might sound the same on one system, different on another. If you hear differences, great. Just don't look down on others who don't hear differences because you have "golden ears".

Error correction
Let's get this one out of the way quickly. One theory a lot of people like to trot out is that differences in discs can be due to reflectivity, how well the discs are burned, pinholes, etc. If not burned well, the error correction has to "work harder" and thus the disc won't sound as good. Gold discs are better, because they have more even surfaces and no pinholes.

This is nothing but nonsense.

CIRC (Cross Interleaved Read-Solomon Code) encoding utilized on CDs can correct gaps up to 2.4mm (3,500 bits) in length. Mind you, that doesn't mean "just make it so there isn't a pop", it means "even with a 2.4mm hole the data is still 100% accurate and intact". In the world of pinholes and scratches, 2.4mm is pretty huge. So unless you have a CD in *really* bad shape, chances are you're going to be getting 100% of the data that's on the CD. Quote:

Another suggestion is that sonic degradation is the result of
uncorrectable data errors on replay. This seems initially unlikely, since uncorrectable errors are very infrequent (at least as evidenced by the VALID flag at the player's output), and the sonic degradation described by listeners is continuous in character.


It's in the mastering
If by mastering you mean "the process of playing back an analog tape, converting to digital, and using EQ/compression/etc. to create a master tape", umm, no. Digitized music is a series of numbers. Total silence could be considered 0. A peak at 0dB could be considered 2^16 (or 2^24 if using 24 bits, for example). On a stereo CD, there are 88,200 16 bit numbers per second. Change the mastering and you change those numbers. If the numbers are the same, so is the mastering. There isn't some hidden 5th dimension where the voodoo is stored, at least not in terms of the data itself.

It's jitter
Very unlikely. Jitter is a time based error - instead of sampling at precise intervals, some samples might come slightly before or after they are supposed to. Mind you, this *can* be a problem with DACs. But they key here is CD players don't get timing data off of the CDs themselves. Quote:

A common theory is that physical imperfections in the 'pits' of the CD cause timing variations in the recovery of the EFM data, which are carried through to the player's digital-to-analogue converter (DAC). This 'sampling jitter' could cause significant audible degradation, as has been documented in [2], [3].

However, this theory misunderstands the basic architecture of the player: the timing clock of the DAC does not commonly depend on the recovered EFM data timing, but is derived from a free-running and stable crystal oscillator. The data is buffered in memory between the disc-reading electronics and the DAC. This buffer is emptied at the crystal oscillator rate, and kept half-full by varying the rotation speed of the disc; therefore
jitter in the recovered EFM data should not affect the clocking of the DAC.


Hell, even someone who hears differences with green pens couldn't find any changes in jitter:

The jitter measurements that I carried out raised more questions than they answered. As with error rates, none of the devices or treatments reduced jitter in the HF signal. During the testing, I had no idea how much jitter reduction to anticipate, and consequently set the measurement parameters so that even the smallest change would be detected. When I measured no change in jitter, I thought that perhaps the jitter analyzer, or my interpretation of the results, lacked sufficient resolution. However, after completing the jitter measurements, I received the Esoteric PD2, a $4000 CD transport. It had enormously lower jitter than the Magnavox player that I had used in conjunction with the error analyzer. Any concerns about the jitter analyzer's resolution were thus allayed. The PD2's jitter reduction was an order of magnitude greater than what I had been looking for during the tweak-effect measurements.


Side note - this answers a mystery I've always wondered about:

"This could be attributed to the fact that PDO uses a slightly different process in creating the center hole. Instead of molding the disc with the hole, it is punched after the disc has been made."

Anyway...

That leaves?
The leading theory (*only* theory?) seems to be this:

Measurements have confirmed that amplitude modulation of the analogue outputs of many one-box CD players by motor and servo-related interference occur. Furthermore, the resulting spuriae are of a character and at a level which would be consistent with noticeable sound degradation for a critical listener. The disc-dependent modulations are low-frequency, and so produce distortion sidebands close to the stimulus frequency. Masking theory suggests that these would be inaudible. The track-position-dependent modulations are generally higher in frequency and amplitude and, as such, may be noticeable to a critical listener. These effects have not been identified in two-box players, which is not surprising since they appear to be caused by modulation of the reference voltage of the internal DAC by the servo and motor electronics.


If I'm reading all of that correctly, it seems to suggest that physical differences between discs can cause the motor and servo to behave slightly differently, which cause voltage differences in the DAC which then result in audible differences. Now, this doesn't explain everything (notably that the people in the blind tests conducted, albeit a *very* small sample, couldn't accurately differentiate between discs), but it goes farther than just about anything else.

Of course, this all points to two things I've been saying for years:

It's in the player
The study I've referenced firmly suggests that different players may handle discs differently. Some may have problems with "bad" discs, while others (notably those with external DACs) most certainly would not. If you can't hear a difference on your player, it might just mean there *isn't* a difference on your player.

Burning/copying will eliminate physical differences
Let's assume that two discs are digitally identical, but yet have different physical properties. Let's assume they sound different on a certain player. Ok. Well, like it or not, if you copy both of those to your PC, those physical differences VANISH. You're left with nothing but the raw digital data. Play them both back on your computer and you won't hear a difference. Burn them both to CD-R, and (assuming you don't change anything in the burning process) they will both sound the same. Now, will they sound better or worse than the original discs? Who knows?! If the original discs had "bad" physical characteristics, it's quite possible that a CD-R could sound better. Or if the originals were "good" and your CD burner isn't so hot, the CD-R could sound worse. Whatever the case, the data hasn't changed. The physical differences only exist at the last step in the chain, and are eliminated each time a copy is made. Let's say you burn a CD with "bad" characteristics and send it to your friend. He burns another "bad" CD and so on. 100 generations of "bad" copies later, somebody gets it and burns it on a "good" CD burner. Assuming there were no digital errors along the way (unlikely), that CD will sound identical to a first generation "good" burn. The process isn't additive.

I think that's about all for now. I'm sure I'll think of various points later, but in the mean time, discuss and share with all of your friends.
"I know because it is impossible for a tape to hold the compression levels of these treble boosted MFSL's like Something/Anything. The metal particulate on the tape would shatter and all you'd hear is distortion if even that." - VD

Andreas
Posts: 535
Joined: Wed Jan 07, 2004 2:41 am

Postby Andreas » Sat Mar 18, 2006 12:11 pm

Here is my theory, that totally contradicts to what Luke wrote above:

Another suggestion is that sonic degradation is the result of
uncorrectable data errors on replay. This seems initially unlikely, since uncorrectable errors are very infrequent (at least as evidenced by the VALID flag at the player's output), and the sonic degradation described by listeners is continuous in character.

Not if the uncorrectable data errors occur at a continuous rate and force the CD player to interpolate frequently.

A burned CDR has more physical errors than a factory-pressed CD. The faster the burning speed, the more physical errors. If you extract the CD with a program like EAC, these physical errors can be corrected accurately, since it reads each stream twice and has enough time to slow down the reading speed. Therefore, the burned CDR will still be bit-identical to the original.

However, a CD player does not have the possibility to slow down if necessary or to read error-laden parts twice. If the errors add up and can't be corrected by the Reed-Solomon-Cross-Interleaved Code anymore, the CD player will use interpolation to avoid skipping or stalling. This interpolation is audible, and it can affect things like the definition of the bass, the clarity of the highs, the soundstage or transients. Therefore, the CDR can sound different from the original CD, even if it is bit-identical.

However, that difference would depend on the CD player as much as on the CDR.

User avatar
Rspaight
Posts: 4386
Joined: Wed Apr 30, 2003 10:48 am
Location: The Reality-Based Community
Contact:

Postby Rspaight » Sat Mar 18, 2006 12:20 pm

Nice summary.

I think it's best if we proceed under the assumption that we are indeed getting accurate rips, and leave out the whole "anything other than EAC" cached-data read error hoodoo voodoo.

For me, a lot of these "error correction/bad bits" fantasies don't pass the smell test because people always assign certain sonic characteristics to these "problems." ("Bad bass," "fuzzy highs," the inevitable "smeared transients.") If the numbers are wrong, it's not going to produce consistent effects. As you accurately point out, the numbers can only be "wrong" across a read error of multiple millimeters, so you'd be talking about pops, clicks, skips and radically "wrong" audio in isolated spots, not subtle sound quality changes across an entire disc.

Remember, kids, each multiple increase in speed equals .25dB reduction at 11kHz. Pass it on.

Ryan
RQOTW: "I'll make sure that our future is defined not by the letters ACLU, but by the letters USA." -- Mitt Romney

User avatar
Crummy Old Label Avatar
Posts: 1226
Joined: Wed Aug 25, 2004 5:55 pm
Location: Out of my fucking mind

Postby Crummy Old Label Avatar » Sat Mar 18, 2006 12:26 pm

Given that the latest trend at SHtv is for some "golden ears" to now "argue" (I use the term quite liberally) that "lossless isn't really lossless," I wonder when they'll start threads about certain brands of harddrives "sounding better" than others?

Two decades and counting of widespread use of digital audio reproduction, yet so few of the self-appointed "golden ears" can demonstrate even a basic grasp of its fundamental workings. Astounding, really and truly.
If you love Hi-REZ TAPE HISS, you're REALLY going to love Stereo Central

User avatar
Rspaight
Posts: 4386
Joined: Wed Apr 30, 2003 10:48 am
Location: The Reality-Based Community
Contact:

Postby Rspaight » Sat Mar 18, 2006 12:29 pm

If the errors add up and can't be corrected by the Reed-Solomon-Cross-Interleaved Code anymore, the CD player will use interpolation to avoid skipping or stalling.


I'm not aware that this has been demonstrated by a transport passing consistently incorrect data from a CD-R. In my experience, CD-Rs are no more likely to require "slow" rips from EAC than pressed CDs.

Ryan
RQOTW: "I'll make sure that our future is defined not by the letters ACLU, but by the letters USA." -- Mitt Romney

User avatar
Rspaight
Posts: 4386
Joined: Wed Apr 30, 2003 10:48 am
Location: The Reality-Based Community
Contact:

Postby Rspaight » Sat Mar 18, 2006 12:32 pm

Crummy Old Label Avatar wrote:I wonder when they'll start threads about certain brands of harddrives "sounding better" than others?


Seagates exaggerate the highs. Maxtors have a bloated midrange. Western Digital boasts a natural yet slightly "sweet" presentation. I listen to music exclusively from Iomega ZIP disks. Completely transparent!!

Ryan
RQOTW: "I'll make sure that our future is defined not by the letters ACLU, but by the letters USA." -- Mitt Romney

User avatar
lukpac
Top Dog and Sellout
Posts: 4591
Joined: Wed Apr 02, 2003 11:51 pm
Location: Madison, WI
Contact:

Postby lukpac » Sat Mar 18, 2006 12:43 pm

Andreas wrote:Not if the uncorrectable data errors occur at a continuous rate and force the CD player to interpolate frequently.

A burned CDR has more physical errors than a factory-pressed CD. The faster the burning speed, the more physical errors.


Any source for this? I don't want to say "I don't believe you", but I don't believe you.

If you extract the CD with a program like EAC, these physical errors can be corrected accurately, since it reads each stream twice and has enough time to slow down the reading speed. Therefore, the burned CDR will still be bit-identical to the original.

However, a CD player does not have the possibility to slow down if necessary or to read error-laden parts twice.


I get bit-accurate rips at higher than 16x extraction. Even with all of EAC's "stuff" turned on (which I often don't use anyway), I always get something faster than 1x overall (more like 4x or 8x or something). Which means the disc is spinning a lot faster than that. Why would a standalone player have more problems at 1x?

If the errors add up and can't be corrected by the Reed-Solomon-Cross-Interleaved Code anymore, the CD player will use interpolation to avoid skipping or stalling. This interpolation is audible, and it can affect things like the definition of the bass, the clarity of the highs, the soundstage or transients. Therefore, the CDR can sound different from the original CD, even if it is bit-identical.


If the errors really "add up" in the way you are describing, I would think there would be far more than would be able to be interpolated, and you'd actually end up getting clicks and such. What are the odds that these errors would just happen to be too large to actually correct but not large enough that they can't be interpolated?
"I know because it is impossible for a tape to hold the compression levels of these treble boosted MFSL's like Something/Anything. The metal particulate on the tape would shatter and all you'd hear is distortion if even that." - VD

Andreas
Posts: 535
Joined: Wed Jan 07, 2004 2:41 am

Postby Andreas » Sat Mar 18, 2006 12:48 pm

Basic question:

If a CD or CDR is full of physical errors (not! those caused by scratches), and if the CD player is forced to use interpolation frequently because of too many errors, could it produce things that are perceived as "bad bass," "fuzzy highs" or the inevitable "smeared transients"?

And please don't give me the "It has not been demonstrated in a blind test" answer.

Andreas
Posts: 535
Joined: Wed Jan 07, 2004 2:41 am

Postby Andreas » Sat Mar 18, 2006 1:30 pm

lukpac wrote:Any source for this?

It's a theory. A theory has no source, because then it would be a fact.

Andreas
Posts: 535
Joined: Wed Jan 07, 2004 2:41 am

Postby Andreas » Sat Mar 18, 2006 1:31 pm

lukpac wrote:If the errors really "add up" in the way you are describing, I would think there would be far more than would be able to be interpolated, and you'd actually end up getting clicks and such.


Any source for that?

User avatar
Xenu
Sellout
Posts: 2209
Joined: Mon Apr 07, 2003 8:15 pm

Postby Xenu » Sat Mar 18, 2006 1:55 pm

Andreas wrote something like to this me...Andreas, I responded to that over at sh.tv, but I'll follow it up here, removing the reundant portions that others have already covered.

Firstly, what Luke said. CDRs don't have more errors, things don't need to slow down, etc. It doesn't happen.

And no, it couldn't. Interpolation doesn't sound like "bad bass." The closest I could give you is "smeared transients," assuming we mean things like instantaneous attack, for which a few fucked up samples could conceivably changed the sound. But it wouldn't be constant.

I have a fucked up CD like that. It's a radio station promo of Tom Petty. It doesn't skip or anything, but the player is forced to constantly interpolate thanks to errors. You know what it sounds like? Fuzz. Lots and lots of white noise.

And responding to your subsequent PM:

But people like Roger Nichols and even, *gasp*, Steve Hoffman have said that they heard differences between the glass master and "inaccurately" pressed CDs. Just because Grant and Dave are ignorant about the science, does not mean that the phenomenon does not exist.


I'd like to respond to two points here.

Just because Roger Nichols is a mastering engineer doesn't mean he's immune to suggestion. We're ALL prone to suggestion...that's the point. I don't deny that people hear differences; I deny that the differences are actually there. Big difference.

(Nor am I positing that I'm immune to this. I hear all sorts of weird shit. But I recognize that it's ME, it's ME it's ME. I'm prone, for example, to suggestion based on how a disc looks...it's hard to explain, but I have a preconceived bias of what I'm "hearing" based on the disc itself. I also "imagine" the sound happening in a space inspired by the artwork. But this is me. I'm not going to describe OK Computer as having a "white" sound because the artwork inspires that as a reaction)

Why am I even interested in this topic? Because I have CDRs that are affected by digital noises. These are not skips or glitches, and if extracted to the PC, the files are clean. How do you explain that?


I explain this in several ways:

a) Are these pre-ripping errors on every player? My old Sony player hated certain CDRs...*hated* them. Mostly the blue Verbatim type Luke really likes. It would made all sorts of awful staticky noises on those discs. No errors, no scratches, no skips, and of course things would be fine on extraction (Extraction being Audiocatalyst in Burst mode).

b) Maybe the CDR was written poorly? It happens. I have one CDR that wouldn't play in ANYTHING, but after a night of serious EACing ending up ripping perfectly (with matching CRCs, to boot). So it happens. It happens to pressed discs too. It may very well be more LIKELY at high speeds, but it's certainly not endemic to them.

I mean, let's face facts here: this is all a scam in the line of other audiophilia. If people didn't believe the crazy voodoo put out there about CDRs, how the fuck would MoFi sell any? MoFi CDRs are *completely unnecessary*, as are Mitsui gold discs, Taiyo Yuden, etc. But MoFi is that much more completely unnecessary, because those discs are EXPENSIVE, whereas TY are a "why not?" kind of purchase. Before college, I bought 200 KHypermedia discs...CMC, "garbage material," etc. FIVE YEARS LATER, not one has gone bad.
-------------
"Fuckin' Koreans" - Reno 911

Andreas
Posts: 535
Joined: Wed Jan 07, 2004 2:41 am

Postby Andreas » Sat Mar 18, 2006 2:38 pm

About the bad CDR...

What does "written poorly" mean here? I am talking about a CDR that I have burned myself at 52 times speed. EAC extracts it at about 14 times speed, no errors. Yes, static noises are more like it, mostly during the first two tracks. No, they are not on every player, my DVD drive reads them perfectly.

How do these noises occur, in your opinion? Following Luke's post, a CD player can only do three things:
1. Correct physical errors and play perfectly, or
2. Find uncorrectable errors and skip, or
3. Suffer from jitter (which this is definitely not!)
That does not explain the behaviour of "static noises".

User avatar
Xenu
Sellout
Posts: 2209
Joined: Mon Apr 07, 2003 8:15 pm

Postby Xenu » Sat Mar 18, 2006 2:47 pm

No, it can also find uncorrectable errors and attempt to interpolate.

Skipping is a PHYSICAL phenomenon, usually. A disc with a perfect playing surface, but bad data (or data that a player doesn't like) won't necessarily skip, but it'll start to interpolate. And when you have thousands of samples in a row that need interpolation, you get static.

If your DVD drive reads it perfectly, you've answered your question. Whatever your burner is doing at 52X (could it be Zone-CLV?), some of your players don't like. Or it's the combination of your burner and the media.
Last edited by Xenu on Sat Mar 18, 2006 2:57 pm, edited 1 time in total.
-------------

"Fuckin' Koreans" - Reno 911

Andreas
Posts: 535
Joined: Wed Jan 07, 2004 2:41 am

Postby Andreas » Sat Mar 18, 2006 2:56 pm

But it's still a CDR that sounds different from the original on a certain CD player. Isn't that the point of the dicussion?

User avatar
lukpac
Top Dog and Sellout
Posts: 4591
Joined: Wed Apr 02, 2003 11:51 pm
Location: Madison, WI
Contact:

Postby lukpac » Sat Mar 18, 2006 3:03 pm

Andreas wrote:What does "written poorly" mean here? I am talking about a CDR that I have burned myself at 52 times speed. EAC extracts it at about 14 times speed, no errors. Yes, static noises are more like it, mostly during the first two tracks. No, they are not on every player, my DVD drive reads them perfectly.

How do these noises occur, in your opinion? Following Luke's post, a CD player can only do three things:
1. Correct physical errors and play perfectly, or
2. Find uncorrectable errors and skip, or
3. Suffer from jitter (which this is definitely not!)
That does not explain the behaviour of "static noises".


It seems to me "static" would be covered under #2, would it not? It can't correct the error or interpolate, so it produces garbage instead.

I've had a few discs over the years that either won't play, or have issues like those described above. They have been few and far between, though, and only with CD-Rs. Are you trying to say a gold disc (UD2) could have so many errors as to cause this problem?
"I know because it is impossible for a tape to hold the compression levels of these treble boosted MFSL's like Something/Anything. The metal particulate on the tape would shatter and all you'd hear is distortion if even that." - VD