Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: CD's weak error correction - artefacts on bought CD's? (Read 11681 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

CD's weak error correction - artefacts on bought CD's?

The shortcomings of CD's weak error correction are now (finally) quite widely-acknowledged. "Ripping" the CD tracks digitally does not automatically give a perfect copy, and a comparison of checksums with those stored in the "AccurateRip" database is the closest thing we have to a guarantee of no ripping errors.

This makes me wonder about commercial compilation CD's. It seems logical to assume that a relatively small record label might simply rip individual tracks from other CD's, to use as the source for their own compilation discs.
If this does occur, and they do not use "secure" ripping software, the tracks on the new disc will be sourced from an inaccurate rip.

The same issue might occur when a CD album is being pressed in a second plant, if a copy of the CD from the first plant was used as the mastering source.

I certainly have numerous poor-sounding CD's in my own collection, to which the above could apply.

Anyone with commercial CD mastering experience care to comment?

CD's weak error correction - artefacts on bought CD's?

Reply #1
Expand on "poor-sounding" - what do you mean, exactly?

Yes, the Red Book standard allows CD players to interpolate new samples on-the-fly if the original data can't be recovered, but to have so many consecutive unrecoverable errors that said interpolation causes an actual audible change in the overall tonal balance of a CD - and yet without any skips - seems highly unlikely.

I think what you're referring to are simply sonic choices made in the tracking, mixing and/or mastering stages of certain CDs by the artists/producers/engineers that you don't like - I have several CDs in my collection that I feel the same way about.
"Not sure what the question is, but the answer is probably no."

CD's weak error correction - artefacts on bought CD's?

Reply #2
CD players do interpolate, that is true.

However, what is being suggested here is a case of reading the original data digitally, with errors, and these errors being encoded into a new copy.

The player will not interpolate these since this is no longer a read error that it detects. In this case there could conceivably be pops etc.

CD's weak error correction - artefacts on bought CD's?

Reply #3
CD players do interpolate, that is true.
However, what is being suggested here is a case of reading the original data digitally, with errors, and these errors being encoded into a new copy.
The player will not interpolate these since this is no longer a read error that it detects. In this case there could conceivably be pops etc.

Why would there be pops?
Pops from an inaccurate rip could conceivably be encoded into a new copy, however these would almost certainly be noticed by the engineers.
The more relevant issue is that of CD's error "hiding". Whereas pops are easy to spot, missing samples replaced by interpolated ones during an inaccurate rip would not be directly noticible or detectable to an engineer doing a quality check, in the absence of reference checksums.

CD's weak error correction - artefacts on bought CD's?

Reply #4
Yes!  This happened to me.  I bought a K-Tel 60's compilation from CD universe that had skips & clicks n some tracks.  The errors seemed to be "embedded" into the data, since the errors were apparent when the CD was played, and on the rip, and EAC didn't show any errors (like pdq was talking about).

The most obvious error was that the track listing on the packaging didn't exactly match the tracks on the CD!!!!  I returned it for a refund.

Of course these were originally analog tape recordings from the 1960's, but the defects sounded like "digital" defects, not tape or vinyl type defects.

CD's weak error correction - artefacts on bought CD's?

Reply #5
Expand on "poor-sounding" - what do you mean, exactly?

Well... I'd rather stick to the scientific side of things, rather than falling into the "audiophile trap" of trying to put personal perceptions and feelings about a given audio source into words.

Yes, the Red Book standard allows CD players to interpolate new samples on-the-fly if the original data can't be recovered, but to have so many consecutive unrecoverable errors that said interpolation causes an actual audible change in the overall tonal balance of a CD - and yet without any skips - seems highly unlikely.

I'd have to disagree here.
However, regardless of whether you believe that the insertion of interpolated samples to replace unrecovered ones creates an audible change, my original question remains valid, and the answer is a simple yes or no.

CD's weak error correction - artefacts on bought CD's?

Reply #6
If you think that inexact CD duplication due to error can result in a change in tonal balance, I'd love to hear some samples.

It simply doesn't work this way.

CD's weak error correction - artefacts on bought CD's?

Reply #7
Yes!  This happened to me.  I bought a K-Tel 60's compilation from CD universe that had skips & clicks n some tracks.  The errors seemed to be "embedded" into the data, since the errors were apparent when the CD was played, and on the rip, and EAC didn't show any errors

Yes, this is one of the shortcomings of EAC and the AccurateRip database - pre-existing errors actually _encoded_ onto a CD by the mastering engineers will not be detected, including any derived from a previous inaccurate rip(!).
Another problem is that of inaccurately-ripped CDs being distributed in lossless formats on file sharing sites - the checksums derived from them are apparently being entered into the AccurateRip database by other internet users, when of course they shouldn't be.

CD's weak error correction - artefacts on bought CD's?

Reply #8
The vast majority of read errors are corrected to exactly what they are supposed to be, not to some interpolated guess. The more extreme types of error correction are required occasionally, but not from the majority of CDs. Even these corrections are rarely audible.

Exact, bit for bit, extraction is common. It isn't so hard, merely tedious, to demonstrate that one can do this on one's own computer. I have. You write a CD-R, with any audio tracks, then you extract from your own CD-R. Compare each extracted track to that from which you wrote it, by one of the methods that will tell you, bit for bit, if there are any differences in the audio data. If variations isn't extremely rare to non-existent, you are doing something wrong or have defective equipment.

Physical damaged CDs are another thing, but flawless playback can still occur in spite of quite a bit of minor scraping and scratching. CD created in such a way as to inhibit copying are a more difficult case and those CD-Rs written at very high speed on mass duplicators can sometimes give one fits.

Those "fits" are usually very obvious and not the least bit musical.

CD's weak error correction - artefacts on bought CD's?

Reply #9
The more relevant issue is that of CD's error "hiding". Whereas pops are easy to spot, missing samples replaced by interpolated ones during an inaccurate rip would not be directly noticible or detectable to an engineer doing a quality check, in the absence of reference checksums.

Surely, if such phenomena were responsible for your “numerous poor-sounding CD's”, they’d be even more blatant to an engineer? Of course, this assumes they’re qualified in hearing, not just nominally and by knowing the right people[/cynic]

Then again, I wonder how many engineers would consider this possibility enough to do a subjective “quality check” but not to give it that little bit more thought whence they would want an objective method with which to verify the accuracy of the copy.

CD's weak error correction - artefacts on bought CD's?

Reply #10
The vast majority of read errors are corrected to exactly what they are supposed to be, not to some interpolated guess. The more extreme types of error correction are required occasionally, but not from the majority of CDs. Even these corrections are rarely audible.

Exact, bit for bit, extraction is common. It isn't so hard, merely tedious, to demonstrate that one can do this on one's own computer. I have. You write a CD-R, with any audio tracks, then you extract from your own CD-R. Compare each extracted track to that from which you wrote it, by one of the methods that will tell you, bit for bit, if there are any differences in the audio data. If variations isn't extremely rare to non-existent, you are doing something wrong or have defective equipment.

Physical damaged CDs are another thing, but flawless playback can still occur in spite of quite a bit of minor scraping and scratching. CD created in such a way as to inhibit copying are a more difficult case and those CD-Rs written at very high speed on mass duplicators can sometimes give one fits.

Those "fits" are usually very obvious and not the least bit musical.

Are these claims based solely on the results of your own experiments on your own computer?

CD's weak error correction - artefacts on bought CD's?

Reply #11
Surely, if such phenomena were responsible for your “numerous poor-sounding CD's”, they’d be even more blatant to an engineer? Of course, this assumes they’re qualified in hearing, not just nominally and by knowing the right people[/cynic]

Then again, I wonder how many engineers would consider this possibility enough to do a subjective “quality check” but not to give it that little bit more thought whence they would want an objective method with which to verify the accuracy of the copy.

Engineers are quite willing to act as enablers for the "CD loudness war", sometimes introducing a huge number of hard clips throughout each track in the process, so I personally don't put too much faith in them.
Either way though, if they were limited to ripping other CD's as an audio source, and didn't have the hardware/software required to make secure rips, then the issue would be largely beyond their control.

CD's weak error correction - artefacts on bought CD's?

Reply #12
Yes, this is one of the shortcomings of EAC and the AccurateRip database - pre-existing errors actually _encoded_ onto a CD by the mastering engineers will not be detected, including any derived from a previous inaccurate rip(!).
Another problem is that of inaccurately-ripped CDs being distributed in lossless formats on file sharing sites - the checksums derived from them are apparently being entered into the AccurateRip database by other internet users, when of course they shouldn't be.


I am sorry to dissapoint you - but Accuraterip only ensures that CD is ripped without errors. What is on that CD is of no concern to Accuraterip database, as long as it is original CD.
It is not a problem of AR database, nor EAC, it is solely the problem with sourcing the CD.
I guess you never listened to glitch kind of music, eh...?

About this another problem, I don't know how AR populates it's database, but IIRC there have to be some number of same submissions for CD to enter database. I may be wrong, though.
Error 404; signature server not available.

CD's weak error correction - artefacts on bought CD's?

Reply #13
IIRC, a single entry will be shown as a match/non-match, but if there are two entries, and those don't agree, then neither is used until there is a third entry that verifies one of the first two.


CD's weak error correction - artefacts on bought CD's?

Reply #15
I am sorry to dissapoint you - but Accuraterip only ensures that CD is ripped without errors. What is on that CD is of no concern to Accuraterip database, as long as it is original CD.
It is not a problem of AR database, nor EAC, it is solely the problem with sourcing the CD.

Err... I never claimed otherwise??

About this another problem, I don't know how AR populates it's database, but IIRC there have to be some number of same submissions for CD to enter database. I may be wrong, though.

You mean the problem of inaccurate rips from file sharing sites? Several people will download the same files, and so several people will potentially submit the same checksums. These "bogus" checksums shouldn't affect any "legal" rippers, however.

CD's weak error correction - artefacts on bought CD's?

Reply #16
Not that it's been discussed to death or anything, but yes, there are no 100% guarantees with AccurateRip and/or secure ripping, though I'm not really sure this is all that pertinent to do with the topic at hand.

CD's weak error correction - artefacts on bought CD's?

Reply #17
Not that it's been discussed to death or anything, but yes, there are no 100% guarantees with AccurateRip and/or secure ripping, though I'm not really sure this is all that pertinent to do with the topic at hand.

It's not :-)
I was more interested in the methods and formats used for distribution of digital recordings to the various CD mastering plants, and to smaller record labels making compilations etc; not just in the present day, but throughout the CD format's history.

CD's weak error correction - artefacts on bought CD's?

Reply #18
Quote
Are these claims based solely on the results of your own experiments on your own computer?


If you are interested, you can read about the error correction technology built into the audio CD standard. This is available many places. While it is often pointed out that it isn't as rigorous as that used for data disks, it might be enlightening to know what this really means. Here are a couple of quotes from a respected technical source

"If within tolerances, errors can be detected and corrected with absolute fidelity to the original data, ..."

"With proper design, digital audio systems such as CD and DVD can approach the computer industry standard, which specifies and error rate of 10 to the minus 12 -- that is, less than one uncorrectable error in ... one trillion bits."

I think that comes out to one uncorrectable error in around 195 hours of CD audio.

As with a data CD, any errors than can be corrected exactly are. What comes out the end is exactly what was intended when the CD was created. Fortunately this is the vast majority of errors, the only errors on most in-good-condition CDs.

Where it gets looser is with more extreme errors. When dealing with non-audio digital data, when an error can't be corrected exactly, you are usually not going to get anything from the computer. However, the listening public isn't oriented to 'perfect or not at all,' they just want to hear the music. So the next step of error correction makes up a little, using the last good sample or what seems likely from the surrounding data. It is probably often correct but there isn't any guarantee. What is practically guaranteed is that it will be so close to the original that you are extremely unlikely to hear anything untoward.

When things get really extreme, and where the computer would issue serious failure message with data, audio CD players just keeps plugging along. Now it may well sound different, perhaps really terrible, like a loud discharge of static electricity, a really big click on an LP, etc. Or, for a single uncorrectable, it may pass unnoticed.

The above has nothing to do with conclusions based on using my PC, it is the way audio CDs systems are designed and built. I freely admit I don't have all the details in my head. I've read about them, I came to accept the terms, I never tried to memorize it all.

The part about writing to CD-R and extracting from CD-R, and getting bit for bit perfect results, has been done on my PCs. It has also been done by many other people. It does not depend on some external standard database or some "secure" DAE checking technology. If you write your own, you have the ultimate standard right there, the source, to check the extraction against.

There are various PC CD read test utilities (e.g. K2probe and CDspeed). The ultimate testers, used in professional installations, are very expensive, but many PC optical drives can report read errors, by type, particularly the first two levels, and free or inexpensive software can collect and interpret these drive messages. Using them on many CDs, I've observed that most CDs have only correctable errors. This will not apply to those on deteriorating medium. And, while badly scratched/scuffed CDs might still have a very good data layer, that data isn't readily accessible;  some people routinely let their CDs come to all sorts of harm and collect all sorts of prints and dirt. But none of these problems are a defect in basic audio CD technology.

I have extracted far less than optimum CDs many times. I've heard the awful crack crack when serious defects, interfere with reading the data. Most often these have been spoken audio, such as audio books. I suspect the cause is most often some form of copy protection or is poorly written CD-R made from marginal media on a high speed copier. It has not been a problem with many music CDs.

I've also listened carefully to many tracks that EAC reported as less than perfectly extracted. I usually just do test and copy, then compare the checksums. Some have had obvious audio defects, in the form of noise, not subtly altered audio. I've not been able to detect anything amiss in the majority.

CD's weak error correction - artefacts on bought CD's?

Reply #19
If you are interested, you can read about the error correction technology built into the audio CD standard. This is available many places. While it is often pointed out that it isn't as rigorous as that used for data disks, it might be enlightening to know what this really means. Here are a couple of quotes from a respected technical source...

Oh, I've read lots about the error correction technology built into the audio CD standard.
I just checked with a _very_ respected technical source, and I'm afraid he disagrees with the claims made by your respected technical source. Here are some quotes:
"Read errors and interpolated samples occur frequently during normal CD playback - far more so than is implied here."

"The claims made here are both vague and unsubstantiated. They have no basis in scientific fact."


No-one's denying that a bit-perfect CD audio rip is possible with a computer's optical drive. The issue is that a perfect rip isn't always _guaranteed_.
Like you, I've gotten bit-perfect copies using insecure ripping modes (the WAV checksums match after manually correcting for the drive offsets), but only from CD's and CD-R's in mint condition. Also, because this works quite consistently on my own optical drive, doesn't mean it will work on all of them.
CD-ROMs, by contrast, contain an extra layer of error correction, which reduces the useable disc storage capacity by approx. 100MB. The format is designed so that the information is either extracted bit-perfectly or not at all.
A modern internet-based analogy would be UDP vs TCP - the former has weaker error correction, but less overhead as a result. Bit perfection is not guaranteed with UDP. It's generally only used for streaming audio and video, but the difference between it and Audio CD, which is carrying a plain old PCM stream, is that the formats typically sent over UDP actually have their own limited error correction built into them (the MPEG2 "transport stream" is an example of this).

 

CD's weak error correction - artefacts on bought CD's?

Reply #20
the difference between it and Audio CD, which is carrying a plain old PCM stream, is that the formats typically sent over UDP actually have their own limited error correction built into them (the MPEG2 "transport stream" is an example of this).

Are you suggesting that CDDA doesn't have built-in limited error correction?  Perhaps you should rephrase.

I just checked with a _very_ respected technical source, and I'm afraid he disagrees with the claims made by your respected technical source. Here are some quotes:

Perhaps your "_very_ respected technical source" can provide us with the basis in scientific fact based for _your_ parroted claims, otherwise you're wasting our time.

I would also like to see the scientific basis behind your reasoning regarding this statement from your initial post:
I certainly have numerous poor-sounding CD's in my own collection, to which the above could apply.

Besides all this, what is the purpose of this thread other than to be argumentative?

CD's weak error correction - artefacts on bought CD's?

Reply #21
Are you suggesting that CDDA doesn't have built-in limited error correction?  Perhaps you should rephrase.

There are two different "levels" (I use the term loosely) being discussed here. CD Audio format has limited error correction. The simple PCM audio stream it carries does not.

Perhaps your "_very_ respected technical source" can provide us with the basis in scientific fact based for _your_ parroted claims, otherwise you're wasting our time.

Why haven't you asked the previous poster to do the same with his "respected technical source".

I would also like to see the scientific basis behind your reasoning regarding this statement from your initial post:
I certainly have numerous poor-sounding CD's in my own collection, to which the above could apply.

Such a discussion would be a diversion from the question I asked, to which the answer remains a simple yes or no.

Besides all this, what is the purpose of this thread other than to be argumentative?

To ascertain whether the potential exists for interpolated samples (created by a widely-recognised shortcoming of the CD audio format) to be erroneously encoded onto newly-mastered commercial discs.
Clearly no-one here has professional mastering experience, so I've wasted my time.

CD's weak error correction - artefacts on bought CD's?

Reply #22
Why haven't you asked the previous poster to do the same with his "respected technical source".
There was nothing inherently wrong in the claims the previous poster made.  Perhaps you (or the guy who is doing your thinking for you) should read them more carefully.

Such a discussion would be a diversion from the question I asked, to which the answer remains a simple yes or no.

The only other question you wrote that wasn't answered is:
Anyone with commercial CD mastering experience care to comment?

Clearly no-one here has professional mastering experience, so I've wasted my time.
Actually there are people here with such experience, though I suppose they don't think this is worth _their_ time, especially considering that you can't manage to present your concerns without also violating TOS #8.

CD's weak error correction - artefacts on bought CD's?

Reply #23
Besides all this, what is the purpose of this thread other than to be argumentative?

To ascertain whether the potential exists for interpolated samples (created by a widely-recognised shortcoming of the CD audio format) to be erroneously encoded onto newly-mastered commercial discs.
Clearly no-one here has professional mastering experience, so I've wasted my time.

How do you expect these few interpolatred samples can change the sound of a title?
I ripped many CDs over time with EAC and decent drives in the last ~10 years. It is nearly impossible to have bad reads. If a cd is really bad scratched there normaly is a click to hear. Even on CDs in bad condition where several ripping errors occured that were interpolated (not AR verified) the music content differs in only a few milliseconds out of a whole song to a errorfree rip. I wonder how that should create a "poor sounding" file!?
This sounds like some bs audiophiles write about the music sounds much better with the correct read offset ripped. I wonder if they believe drive offset is a voltage
Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

CD's weak error correction - artefacts on bought CD's?

Reply #24
Why haven't you asked the previous poster to do the same with his "respected technical source".


Because its not his job to make your posts for you?