Hi Jan,
On the C1 flags, those are corrected as long as they don't pass along to the C2 corrector and they can't be corrected. Once the C2 flag is active, those data words are junk forever. The concealment might be bad, or excellent. Depends on the player.
The C2 flag should be a rare event on new or discs in really good condition. However, once life has a go at them, the C2 flag can be a common occurrence. Any circular scratch generally will make C2 active, but a scratch from center out isn't nearly as bad.
Hi DF96,
-Chris
On the C1 flags, those are corrected as long as they don't pass along to the C2 corrector and they can't be corrected. Once the C2 flag is active, those data words are junk forever. The concealment might be bad, or excellent. Depends on the player.
The C2 flag should be a rare event on new or discs in really good condition. However, once life has a go at them, the C2 flag can be a common occurrence. Any circular scratch generally will make C2 active, but a scratch from center out isn't nearly as bad.
Hi DF96,
I don't think that is a requirement. They only have to be able to gather the encoded data to replace what was damaged. In other words, they must use the Reed-Solomon code. They can take every C2 error and pass it directly on to the DAC. Those machines really do exist sporting the CD logo. Now for the hard part. There are machines that are only 14 bit, and some CD walkabouts are only 12 bit! How bad can it get? Only your imagination limits the depths some manufacturers will sink to. Those are really inexpensive players, and most often portable.
-Chris
yes, it is rare to have an 'uncorrected' error. Very rare. But, so what if it isnt corrected to the original bits (not bit perfect to the original source). 1000's of interpolated or concealment corrections? It works but doesnt sound like the original source, IMO.
Narrowing things down further.... I dont have the same error 'correction' problem/issues or not nearly to the same extent when I have an HD file downloaded to computer memory file from the Internet and played back from memory. Quit different systems in how they handle/process data.
THx-RNMarsh
Narrowing things down further.... I dont have the same error 'correction' problem/issues or not nearly to the same extent when I have an HD file downloaded to computer memory file from the Internet and played back from memory. Quit different systems in how they handle/process data.
THx-RNMarsh
Last edited:
Why do you keep repeating that incorrect statement? Corrected errors are THE ORIGINAL DATA. Interpolated errors are almost zero.
Can I take it that you haven't bothered to learn about how error correction (CIRC) works?
Can I take it that you haven't bothered to learn about how error correction (CIRC) works?
Chris, that's not what that Clover Labs manual seems to say. Or how am I mis-reading it? I have read the same elsewhere, I'm perfectly happy to be wrong.
Another question, if interpolation becomes so frequent on old CD's why have I not seen any AES papers on improved versions considering current SOTA DSP capability.
SY beat me to it. Richard, the corrected errors, the '1000's', are corrected 100% and they are indistinguishable from the originals - they ARE the originals in a very literal sense.
Stop worrying 😉
OT Edit: I was just reading an article from Julia Belluz (Vox, Wash DC) about the fact-less society, while wondering what happened to the meaning of English words. Doesn't 'correction' have a clear meaning? Not any more alas, because the media have given their own slant to words that were once clear and unambiguous. The article also descents into why politicians like Farage, Trump, Clinton, Putin apparently can lie through their teeth and their fans think nothing worse off it. I won't go into details but the subject is very interesting from a culture point of view.
Jan
Last edited:
Why they teach this stuff to the Navy I don't know. Description of CIRC as used on an audio CD. C1 and C2 are separate Reed-Solomon strategies both designed to correct errors exactly.
https://www.usna.edu/Users/math/wdj/_files/documents/reed-sol.htm
George - It's getting hard to find this stuff, so many broken links.
https://www.usna.edu/Users/math/wdj/_files/documents/reed-sol.htm
George - It's getting hard to find this stuff, so many broken links.
Hi Scott,
Once the C2 flag has been active, that data packet is unrecoverable. No fixing it, so you move on to the second phase - concealment. There is a lot of misconception about what is and isn't correctable with music CD data. Computer data error correction is a lot more robust - as we can all appreciate when we are able to install a new program from a scratched disc. Music program material is another thing again. We are stuck with the Reed-Solomon encrypting and there isn't anything more we can do as far as that area in the reproduction chain (inside the CD Player).
But, what does this mean in practical terms? With new media, or media that is in great condition, not a while heck of a lot. If you have a good CD player, it will have a good interpolation routine which will hide most errors. Its the persistent errors coming off the CD that causes audible issues. The average person would not be able to hear a single sample mute. The analog filter following the D/A would blend multiple errors nicely. If you had a cheap CD player, I guess you could see a reason for poor quality sound (like it needs another reason to sound bad).
Hi SY,
Comment to me?
The one really important factor in all of this is the condition of the media. If it is all scratched, then the C2 flag will be much more active. This is telling you that massive amounts of data is corrupted and can't be fixed. This shouldn't come as a surprise to anyone.
So, with a like-new CD that was pressed well, your error rates will be low and you would probably not notice the results of C2 data errors. I have new CDs that weren't pressed well and they light up the C2 flag pretty often. So these C2 errors can be quite common, or rare. It all depends on your CD, and your CD player. Tracking problems will most definitely have an effect on sound quality. So the quality and state of repair your own personal CD player is in does directly affect how good the CD is going to sound. It is also why a "bit perfect" recording may not be possible in some cases.
-Chris
Once the C2 flag has been active, that data packet is unrecoverable. No fixing it, so you move on to the second phase - concealment. There is a lot of misconception about what is and isn't correctable with music CD data. Computer data error correction is a lot more robust - as we can all appreciate when we are able to install a new program from a scratched disc. Music program material is another thing again. We are stuck with the Reed-Solomon encrypting and there isn't anything more we can do as far as that area in the reproduction chain (inside the CD Player).
But, what does this mean in practical terms? With new media, or media that is in great condition, not a while heck of a lot. If you have a good CD player, it will have a good interpolation routine which will hide most errors. Its the persistent errors coming off the CD that causes audible issues. The average person would not be able to hear a single sample mute. The analog filter following the D/A would blend multiple errors nicely. If you had a cheap CD player, I guess you could see a reason for poor quality sound (like it needs another reason to sound bad).
Hi SY,
Comment to me?
The one really important factor in all of this is the condition of the media. If it is all scratched, then the C2 flag will be much more active. This is telling you that massive amounts of data is corrupted and can't be fixed. This shouldn't come as a surprise to anyone.
So, with a like-new CD that was pressed well, your error rates will be low and you would probably not notice the results of C2 data errors. I have new CDs that weren't pressed well and they light up the C2 flag pretty often. So these C2 errors can be quite common, or rare. It all depends on your CD, and your CD player. Tracking problems will most definitely have an effect on sound quality. So the quality and state of repair your own personal CD player is in does directly affect how good the CD is going to sound. It is also why a "bit perfect" recording may not be possible in some cases.
-Chris
That's not what the professor says unless Jan is right on the liberal use of "correct". One thing we're all missing is that single sample mute can't happen, a single missing sample is always recoverable by RS. I stand corrected the math says e-19 probability.
Last edited:
Hi Scott,
Well, I don't know how you would recover data that doesn't exist anymore. The reconstruction was already attempted in the C1 phase. That's why a C2 error is so rare with media that is in good condition.
We were taught by the folks who went to Japan, and some of the Japanese engineers that came over here to teach the fascinating new world of the CD. The promise of error free play was made back then, but they had to come clean about that. The biggest problem back then was to keep the servos locked to the CD. All of them. Once you lose lock, the C2 comes alive - big time.
Anyway, C1 = error detected and repaired (= original data), C2 = data unrepairable, unrecoverable, gonso. No adjacent data was enough to recreate the missing data. Consider the C2 flag like letting the magic smoke out of an IC package. It's really over for that particular data package. That's when they look around to find stand-ins, look alikes or even just a space. C2 means it's all over. They must be thinking of the C1 flag and it's subdivisions that Mark has brought forward. They hadn't split the definitions that finely when we were taught. Any of the CD seminars I had attended later on didn't make mention of any more than the C 1&2 error flags. What Mark is dealing with is news to me. All I need to know is that a C2 x error has occurred. Now, what did the system do to hide it?
-Chris
Well, I don't know how you would recover data that doesn't exist anymore. The reconstruction was already attempted in the C1 phase. That's why a C2 error is so rare with media that is in good condition.
We were taught by the folks who went to Japan, and some of the Japanese engineers that came over here to teach the fascinating new world of the CD. The promise of error free play was made back then, but they had to come clean about that. The biggest problem back then was to keep the servos locked to the CD. All of them. Once you lose lock, the C2 comes alive - big time.
Anyway, C1 = error detected and repaired (= original data), C2 = data unrepairable, unrecoverable, gonso. No adjacent data was enough to recreate the missing data. Consider the C2 flag like letting the magic smoke out of an IC package. It's really over for that particular data package. That's when they look around to find stand-ins, look alikes or even just a space. C2 means it's all over. They must be thinking of the C1 flag and it's subdivisions that Mark has brought forward. They hadn't split the definitions that finely when we were taught. Any of the CD seminars I had attended later on didn't make mention of any more than the C 1&2 error flags. What Mark is dealing with is news to me. All I need to know is that a C2 x error has occurred. Now, what did the system do to hide it?
-Chris
Last edited:
Chris I certainly respect your input but can't rationalize the two descriptions of how it works. Below is essentially the same as the original AES 1982 paper which also used C1 and C2 as separate Reed-Solomon codes. I don't see the connection between RS and interpolation.
Last edited:
Hi Scott,
I could be wrong, but this may very well explain a lot of confusion.
This is software only. We didn't get down to this level for the simple reason that this stuff is fixed. It can't be broken or repaired as it is hard coded. Repair people don't need to know this stuff at all.
What I'm talking about are hardware test points for error flags. The C1 flag indicates that a "download error" has occurred and that it will attempt to correct the problem. Once the C2 flag activates, this is a test point remember, all attempts at reconstructing the damaged data have failed. It's all up to the concealment stage, whatever the manufacturer has included in that product. The different actions I mentioned earlier can be tried in sequence, the final option in a nice CD Player might be a volume fade to mute if there is a lot of damage but it can still track (really odd, I can't think of any situation that would allow this). Maybe a contrived test they put together, software people can be pretty funny folks!
Anyway, in the case of distorted audio, or "chirping", the test points allow us to determine whether there is a problem in the data reading portion of the machine, or if the problem lies in DSP or clocks somewhere. The C1 flag actually tells us how well the machine is reading the disc (assuming a good disc). We can prove to a customer that it's either the CD itself, or the player. The C2 flag is less active, but it can tell us that the C1 errors occur in groups dense enough to cause the concealment firmware to be active. This is not only rare that we need to use this, but the existence of these test points were not taught anymore, then the test points themselves disappeared too. They probably still exist on pins labeled "factory use only", or something to the same effect.
Anyway, I really think that we are using the same notation for two different things that are unfortunately related to each other. I can see how confusing this could be for someone.
I even questioned why the engineers taught us the Reed - Solomon code. All we had to know was that backup data was distributed in other data blocks in the same area. The actual code does't help a technician troubleshoot a problem at all, just the concept.
Is that helpful?
-Chris
I could be wrong, but this may very well explain a lot of confusion.
.
This is software only. We didn't get down to this level for the simple reason that this stuff is fixed. It can't be broken or repaired as it is hard coded. Repair people don't need to know this stuff at all.
What I'm talking about are hardware test points for error flags. The C1 flag indicates that a "download error" has occurred and that it will attempt to correct the problem. Once the C2 flag activates, this is a test point remember, all attempts at reconstructing the damaged data have failed. It's all up to the concealment stage, whatever the manufacturer has included in that product. The different actions I mentioned earlier can be tried in sequence, the final option in a nice CD Player might be a volume fade to mute if there is a lot of damage but it can still track (really odd, I can't think of any situation that would allow this). Maybe a contrived test they put together, software people can be pretty funny folks!
Anyway, in the case of distorted audio, or "chirping", the test points allow us to determine whether there is a problem in the data reading portion of the machine, or if the problem lies in DSP or clocks somewhere. The C1 flag actually tells us how well the machine is reading the disc (assuming a good disc). We can prove to a customer that it's either the CD itself, or the player. The C2 flag is less active, but it can tell us that the C1 errors occur in groups dense enough to cause the concealment firmware to be active. This is not only rare that we need to use this, but the existence of these test points were not taught anymore, then the test points themselves disappeared too. They probably still exist on pins labeled "factory use only", or something to the same effect.
Anyway, I really think that we are using the same notation for two different things that are unfortunately related to each other. I can see how confusing this could be for someone.
I even questioned why the engineers taught us the Reed - Solomon code. All we had to know was that backup data was distributed in other data blocks in the same area. The actual code does't help a technician troubleshoot a problem at all, just the concept.
Is that helpful?
-Chris
Yes, on the way home I realized simply interpreting the flag as C2 was used vs. C2 failed would cause the same confusion.
Thank you Anatech/Chris. I may not be using the correct words in the strict tech sense used in CD but I do get your message and that is really all I care about.
You have answered my fundamental question(s).
THx-RNMarsh
You have answered my fundamental question(s).
THx-RNMarsh
Last edited:
I had an old badly scratched up CD that wouldn't play on any player or computer CD drive, except one particular old Plextor drive in a PC. Used some software that would try reading it at a specified speed and keep doing read retrys as many times as desired, and log any errors. It was able to extract all the data with no errors, and I burned the files to a new audio CD. Problem solved.
Hi Mark,
Impressive. You weren't going to fail, were you?
Different drives and servo electronics each have problem types they can't deal with and others they do particularly well at coping with. You found the combination that worked for you. Let's just say that your experience is unusual in that the disc would have just as easily been lunched for good. Another disc may confound the Plextor drive. Even the state of alignment matters and that drive that worked may actually be misaligned. But it worked great in this instance.
-Chris
Impressive. You weren't going to fail, were you?
Different drives and servo electronics each have problem types they can't deal with and others they do particularly well at coping with. You found the combination that worked for you. Let's just say that your experience is unusual in that the disc would have just as easily been lunched for good. Another disc may confound the Plextor drive. Even the state of alignment matters and that drive that worked may actually be misaligned. But it worked great in this instance.
-Chris
Hi Richard, Scott,
Thanks. I'm going to bet that this is one of the things that had people arguing bitterly. Myself included.
-Chris
Thanks. I'm going to bet that this is one of the things that had people arguing bitterly. Myself included.
-Chris
No, it is not in software. Each and every digital error correction and signalling is implemented using shift registers with feedback, according to the prime polynomials coefficients defining the correction algorithm. Essentially you parallel load the data in the register, then clock it out and get the right decoded data in a serial format, and the flags for uncorrectable error signalling, if they apply. Of course, this can be multi-stage, etc... But the principle is the same. You may look at the registers feedback as some sort of microcode, but then each and every digital state machine (like a CPU core with its orthogonal instruction set) could be considered "software".
From a designer perspective, this is so standardized it doesn't take more than a few lines in HDL to describe it, and the silicon block is automatically generated at the output.
Nitpicking, of course...
- Status
- Not open for further replies.