[gunnar]

I burn a lot of audio master CDs and they sound much better burned at 1x speed than faster speeds. My new Mac Mini with superdrive will only burn at 8x and 24x. I'm using Jam and Toast 5. Is there any way to force the drive to burn at 1x speed?

[mduell]

Isn't the 1x CD burning thing (that it burns deeper, or makes the audio sound better, or whatever) a myth? I can burn data CDs at 48x that are "perfect" (every bit is as it should be when read back), so why would it make a difference for audio CDs?

[gunnar]

I believe it has to do with the depth and length of the pits. As CD players read the data in real time (and have a hard time of getting it right in the first place), unusual pits tend to introduce error. Yamaha even brought out an audiophile CD burner a while ago to specifically deal with this. Beyond that, you can definitely hear the difference on a good system which is good enough for me.

[tooki]

In theory, it has to do with the shape of the burned "pits". (The laser in no way burns indentations into the dye; it simply changes its color to prevent it from reflecting infrared light.)

In practice, it's nonsense. The error correction in the CD makes it basically impossible for small differences in pit formation to make a difference. This is because the "raw" audio data is encoded using eight-to-fourteen modulation to introduce substantial (75%!) redundancy.

What is true is that burned CDs are never as readable as a real one, so extra margin for error (which is what the Yamaha F1 burner basically did) can't hurt. If a disc is marginal for other reasons, such as cheap media, dirt, scratches, etc, the pit misshapenness could push it over the edge and contribute to errors.

But the problem is, lower burn speeds don't necessarily translate to better burns. On the contrary, some believe that it makes it worse, by prolonging the amount of time the dye is heated, causing distortions.


Anyway, as for your concrete question: if 8x is the slowest you're offered, that's the slowest it knows how to do. Whether it's better or worse to burn fast or slow in that drive is something best tested with proper disc testing methods and equipment.

tooki

[mduell]

Oh, I've heard all sorts of theories: pit depth, pit lengh, jitter, etc. I think it's all bogus, because I can burn an Audio CD from a WAV at 20-48x, rip it at 20-48x, and the two WAV files will be identical.

Sorry I can't help you with the mac mini. Just wanted to know if you had some sources on this 1x theory/myth.

[P]

Originally Posted by tooki

In practice, it's nonsense. The error correction in the CD makes it basically impossible for small differences in pit formation to make a difference. This is because the "raw" audio data is encoded using eight-to-fourteen modulation to introduce substantial (75%!) redundancy.

I seem to remember that it's not (or was not) mandatory for the CD player to support the full error correction code. I'm way too lazy to the math, but I remember that 8-to-11 meant that you could flip one bit and still get the correct value back. I think that 8-to-14 means that you can flip 2 bits and still find the correct value. If so, the CD manufacturer could - at least in theory - implement error correction based only on the first 11 bits and discard the last three unread, because that is still enough to fix single bit errors.

If the CD-player in your computer implements the full 2-bit correction but the car CD-player or home stereo only implements the cheaper 1-bit, that player could be in trouble. Also don't forget that the disc can be damaged later on, introducing more faults.

In general, I think that the 1x thing was more from the days with older burners with crappy lasers. I know that it was a problem back then, that a disc burned at too high speed (this being 4x or 8x at the time) worked fine in a computer but not in a car CD-player (say). I don't know if it's because of the reason I stated above, I'm just making up theories for fun...

[P]

Originally Posted by mduell

Oh, I've heard all sorts of theories: pit depth, pit lengh, jitter, etc. I think it's all bogus, because I can burn an Audio CD from a WAV at 20-48x, rip it at 20-48x, and the two WAV files will be identical.

That doesn't prove anyhing. If you made 100 discs at 48x and 100 discs at 1x, used them for a few years and then read them back, then you might have proven something.

[ghporter]

The 8-14 encoding means you can have something like 6 bits out of 14 faulty and still get the correct data out from each symbol. (It's all about "symbols," not "bytes' because it's just raw data until it's decoded.)

CD encoding and decoding uses two layers of the Reed-Solomon algorithm, which means that the decoder can correct error bursts of up to 4000 bits, and less extensive errors throughout the run of the disc. I've read that the 4000 bits equates to about 2.5mm of disc surface-that's quite a bit of disc!

Reed-Solomon is good enough to get clear, readable data from the Voyager probes over billions of miles through a radio link that STARTS with only 20 Watts of power!

[Chuckit]

Originally Posted by P

That doesn't prove anyhing. If you made 100 discs at 48x and 100 discs at 1x, used them for a few years and then read them back, then you might have proven something.

You're suggesting a fast burn causes the medium to decay faster?

[P]

Originally Posted by Chuckit

You're suggesting a fast burn causes the medium to decay faster?

No. I'm suggesting that a fast burn produces a lower quality disc, a disc with less margin for error (=future external damage).

[tooki]

Originally Posted by P

I seem to remember that it's not (or was not) mandatory for the CD player to support the full error correction code. I'm way too lazy to the math, but I remember that 8-to-11 meant that you could flip one bit and still get the correct value back. I think that 8-to-14 means that you can flip 2 bits and still find the correct value. If so, the CD manufacturer could - at least in theory - implement error correction based only on the first 11 bits and discard the last three unread, because that is still enough to fix single bit errors.

If the CD-player in your computer implements the full 2-bit correction but the car CD-player or home stereo only implements the cheaper 1-bit, that player could be in trouble. Also don't forget that the disc can be damaged later on, introducing more faults.

The 8-to-14 modulation isn't optional, it's the core encoding on the disc. The extra bits of data aren't just tacked onto the end; you have to read the whole code and look it up. (I don't pretend to totally understand it, but wikipedia has a quick explanation of it.)

What I think you're thinking of is interpolation, which is another layer of redundancy. If whole samples are missing, interpolation is used to calculate values to fill in the gaps. Some players' interpolation is better than others', allowing larger gaps to be filled, or better guesses as to how to fill gaps.

And of course the quality of the DAC is critical.

tooki

[ghporter]

R-S encoding is performed very simply through a table lookup. Since there are two layers, it's done through two different tables-this makes it VERY FAST, by the way. As tooki says, it's a process performed on the entirety of the data; the WHOLE bitstream is broken into groups of bits that are processed (twice) to produce the encoded output. And that is why it works so well; since the process is 100% reversible, what you put in is what comes out (within the correction limits of the code). The differences you see in higher and lower quality hardware are more from the buffer that stores data read direct5ly off the CD, and the digital noise filtering (that' the "1 bit" stuff) than anything else.

I once picked up a CD that had been lost out of someone's car and run over, and (being kind of goofy at the time) I took it home. I washed it off to clean off road grime and sand, dried it and played it on my CD player. This was almost ten years ago, and the CD player was about ten years old then-think 1987 vintage hardware. It really did work, as badly mauled as the disc was. THAT is how good this system is.

[P]

Originally Posted by tooki

The 8-to-14 modulation isn't optional, it's the core encoding on the disc. The extra bits of data aren't just tacked onto the end; you have to read the whole code and look it up. (I don't pretend to totally understand it, but wikipedia has a quick explanation of it.)

What I think you're thinking of is interpolation, which is another layer of redundancy. If whole samples are missing, interpolation is used to calculate values to fill in the gaps. Some players' interpolation is better than others', allowing larger gaps to be filled, or better guesses as to how to fill gaps.

And of course the quality of the DAC is critical.

tooki

I know all that. The DISC must have the complete error correction code, but the PLAYER doesn't have to use all of the code.

The point about error-correcting codes in general is that for every legal code X, every code that is a small change from it is illegal and more similar to X than every other legal code. A simple error correction system has a lookup table that translates any illegal code to the closest legal code. The legal codes are then translated into data without the error correction (in practice, these two steps happen at the same time, but it's easier to explain this way).

If you assume that you aren't going to have any errors at all, you can ssume that all codes will be legal and skip the first step - just go directly to the step where you translate the legal code into its non-errorproofed version. Any illegal code is an electronical don't-care value, so the output is unknown. The middle option is to only use parts of the error correction code so you can fix single bit errors. This makes the first table smaller and the electronics simpler, and this is what I think cheaper players (with cheap DACs) do.

This is way OT. I don't really think that this has anything to do with the faults about burning music at a too high speeds. There is something there, however, because I have heard a ticking noise from burned CDs in regular players, CDs that worked fine in a computer, way back in the mid-nineties.

[tooki]

8-to-14 modulation (EFM) isn't error correction code, it's error prevention. (It is not a calculated checksum, as you would find in, say, a RAID system.)

EFM is modulation, whose purpose in this case is to ensure that you never have less than two zeros adjacent, and never more than 10. Why? So that there is a constant stream of 1's and 0's (even if the data were strings of millions of 0's) going to the pickup. It's a way of ensuring that the laser has enough transitions that it can always recalibrate and refocus, but also to prevent the transitions from being too frequent (remember that in a CD, it's the transition from a pit to a land that indicates a 1, not the pit or land itself). Ergo, it is a form of error prevention, not correction. It serves to provide a signal that lends itself to laser pickup.

In playback, the EFM is undone. If some samples are missing, that's OK, because the next layer is the CIRC error correction layer. Then the audio is demultiplexed. After this is where a player might be able to introduce better error correction, by using a better algorithm to replace missing values. Then it's off to the DAC.

All in all, I find it hard to believe that the CIRC error-correction in one player is going to be different enough to make any difference. (And audiophiles' claims of certain errors causing specific frequencies to be affected is... baloney. Errors are equally likely to occur on any bit, so it could affect any frequency.) No, I think the differences are going to be in the pickup system (how good's the optical assembly, how good are the circuits that clean up the raw signal from the laser pickup) and above all from the DACs.

tooki

[P]

I'm sure you're right, tooki, but could you explain this: I know that when you burned discs at a high rate in the late nineties, you got a ticking noise from most players but computers always played it back perfectly. I haven't heard anything like this in years, but I know it was a fact back then. Why is this?