Asus and Sony, perfect C2 ?

[Pio2001]

In http://www.digital-inn.de/showthread...5&pagenumber=4 , 6th post in this page, Registered mentioned that the sony crx220e1 and the asus crw5224 had 100% C2 accuracy.

Is it true ? How does CD Speed 2000 test for C2 accuracy ? What were the exact results ?

Hello Pio2001,

It was me who said that, but to be honest, I didn't do the test.

I just read them at cdr-info.com

These are the relevant links:

http://www.cdspeed2000.com/go.php3?l...errortest.php3

http://www.cdr-info.com/Sections/Art...p+Vol1&index=3

http://www.cdr-info.com/Sections/Art...D%2DRW&index=2

http://www.cdr-info.com/Sections/Art...D%2DRW&index=2

Hope you find them useful.

Best regards.

[JeanLuc]

PIO ... take a look at www.cdrinfo.com/hardware/CDRW 48x ... they have a new test procedure using ABEX Test CD's and unreleased Nero software to determine error correction abilities and C2 accuracy

[Pio2001]

Very interesting... but I don't know if it can be 100 % trusted.

Let me show you why. Here is an old CDR read by a Teac 540e CD ROM drive :



There is a first C2 mistake (blue), much before the rest. Imagine it didn't happen. It's quite possible. In this case, the drive passed the C2 tet with 100% accuracy... until the error rate grew higher than about 400 or 500 mono samples per second.

The Asus tested at CDRinfo (http://www.cdr-info.com/Sections/Art...D%2DRW&index=2) never reached this error rate with the Abex TCD 726. It reached it only for two very short times with the EAC test CD. Only the TCD 721R reached this error rate for a significant time.

How do we know if the drive doesn't miss some errors at higher error rates ?

These tests are very interesting for comparing the DAE ability of the drives, but the C2 results must be taken with caution.

[JeanLuc]

I am still doubtful that it is possible to clearly reproduce specific errors during mastering of CD's ... but IMHO CDRInfo's latest approach is the best public one available to check out a drive's DAE performance, including error correction.

[BobHere]

Originally posted by Pio2001


How do we know if the drive doesn't miss some errors at higher error rates ?

These tests are very interesting for comparing the DAE ability of the drives, but the C2 results must be taken with caution.

Hi Pio,

As you know I have some opinions on C2 accuracy and its relevance.

Two points,

1.
For most of us the high C2 error rates are not relevent. For high C2 error rates EAC is not likley to return an accurate rip and may indeed return a rip that has significant audible artifacts due to hold interpolation. If there is no other source for the CD some people (myself included) may resort to editing the resultant wave to improve its sound. In that case I doubt that inaccurate C2 reporting would make it that much worse and certainly no less accurate.

2.
Unless there is some elementary errors in the drive logic, what the C2 reporting actually shows is where the drive has detected an error (rightly or wrongly) and thus interpolated the signal. This will result in a wrong sample at that location unless the interpolation happens to independently produce the correct value (due to the nature of music there is a relatively high chance of this happening compared to a truly random signal). So from that perspective C2 is always correct. This could be tested by checking each C2 reported sample to see if it is interpolated.

The issue of C2 missing the detection of errors is quite different. In these cases, it is more that CIRC does not detect them due to either its own strategy or some other factors. If CIRC does not detect them they cannot be interpolated and the resultant wave will probably have audible clicks in it. Except for one single CD, I have only seem this happen on CDs with extreme damage where there were serious audible artifacts in the wave produced by large numbers of hold interpolations as well as single sample impulses.

From what I understand about how EAC works, the exact accuracy of C2 reporting is not important where large numbers of errrors occur as EAC relies on its reread process to determine the output wave. All that is necessary is for it to detect at least one error to start the reread process. I believe that C2 reporting should be able to provide this.

The situation is different for Plextools where C2 is more extensively used.

With the current tested performance of C2 reporting on drives equipped with the Mediatec chipsets (the fact that not all drives using these chipsets seem to have the same C2 reporting or CIRC performance tends to show that there are other factors involved in that performance), it is my hope that Andre will include some new C2 based secure modes similar to those used in Plextools so that we can take all of this out of the theoretical situation into a practical situation and do some real testing. I believe that the result will be more accurate rips (on damaged CDs) for some of these late CD drives.

Another addition I would like would be the inclusion of a C2 reporting file into EAC . This woiuld be the C2 information associated with the final written wave showing all interpolated samples. Such a file could be used by post processors to produce better sounding rips from irrepairably damaged CDs where the CD drives have used the very simple interpolation schemes that produce such obvious audible artifacts.

I think those two additions would keep the 'art' moving forward and, as a side benefit, give us plenty of grist for future robust discussions.

Regards,
Bob

[BobHere]

Originally posted by JeanLuc
I am still doubtful that it is possible to clearly reproduce specific errors during mastering of CD's

Hi Jean Luc,
It is quite possible to introduce specific error patterns into a CD during mastering. The hard part is to ensure tha these are the only errors produced or at least that the inherent error rate of the disc is sufficiently low as to keep the test disc valid.

With the advances in mastering, it is possible to produce almost perfect masters. Also injection molding has come a long way and it is also possible to produce very high quality CDs. BUT this all costs money, further each disc would probably have to be tested and certified with a resultant high reject rate.

So if the ABEX test CDs rely on artificial error patterns, perhaps the above reasons may go towards explaining their high costs.

Regards,
Bob

[Halcyon]

This is an interesting discussion.

I've been following the CDR Info measurements as well.

Couple of points I've considered.

1) Their results are relative and can only be compared to other drives they've tested with the exact same methodology (it has changed once or twice during the time they have run the DAE quality / C2 error tests)

2) Pressed C2 errors are only data level errros. Most real life C2 errors are physical level errors. To read these properly the transport, pick up and servo also matter a lot. Even reading the same spot twice might produce unreadable and readable results on the two occasions (this is what EAC uses too).

3) Could we have a STANDARD physical layer error disc (not just data layer) that had scratches of several levels of severity. This would not only test the C2 ability, but also the pickup/servo/transport performance. As such it would be a much better approximation of real life performance with scratched discs.

Comments?

regards,
Halcyon

PS In a way I hope that we are discussing the remnants of a dying standard which will hopefully be supplanted by SACD/DVD-A or something else... Redbook implementation has way too much black magic associated with it to be a contender for a reliable standard in audio delivery in the 21th century, imho.

[JeanLuc]

Originally posted by Halcyon
PS In a way I hope that we are discussing the remnants of a dying standard which will hopefully be supplanted by SACD/DVD-A or something else... Redbook implementation has way too much black magic associated with it to be a contender for a reliable standard in audio delivery in the 21th century, imho.

Hey Hal,

nice to read you once again ... you really think RedBook PCM Audio will die during the next 10 years at its current level of aquaintance ? I surely will not. More likely there will be hybrid discs utilizing both SACD/DVD-A and standard CD for the folks with less money ...

[Halcyon]

JeanLuc,

no, I don't think it'll die.

But I'm hoping it will die regardless of what I believe in

Honestely, it'll take a long time and maybe DVD-/SACD won't be the formats that supplant it.

Nevertheless, I wish it dead soon.

regards,
Halcyon

[JeanLuc]

Uh Huh ... all these new formats come along with DRM ... and improved sound quality as well

[Halcyon]

I know what you mean (DRM).

Still, I'd rather have a real crypto-based DRM that doesn't screw up audio quality (well, DVD-A has optional audio content watermark, but I think it'll die soon enough).

The problem with CDs is that because they are so easy to copy, they will be fucked up by all sorts of nasty copy protections that make them unplayable everywhere and even worsen the sound quality.

I'd rather have limited ripping capability, but GREAT sound quality than limited sound quality and GREAT rippability.

But then again, 99.999% of the music I listen to I own (I think I have 2-6 burned CDs?) and even that is mostly vinyl...

regards,
Halcyon

PS Back to the original thread: does anybody know what the test CDs are that CDR Info uses? I haven't heard of them before...

[JeanLuc]

ABEX Test CD's are highly expensive CD's designed for professional use foremost but not only. However, I do not know if they are available to single customers or end-users ... but you can send a mail at the cdrinfo staff to obtain more details ...

[BobHere]

Originally posted by Halcyon
This is an interesting discussion.

1) Their results are relative and can only be compared to other drives they've tested with the exact same methodology (it has changed once or twice during the time they have run the DAE quality / C2 error tests)

Unfortunately so. The Abex discs probably produce consitent results from test disc to test disc otherwise they wouldn't be much good. The EAC test discs will depend on who made them so you would have to use the exact same one to get directly comparable results. Nevertheless, as long as the EAC discs are not too far apart, the test results should be indicative of performance. I would like CDR-info to run the current tests on some older drives so that we can get direct comparisons between generations of drive. I would be especially interested in testing the whole range of Liteons (due to their populartiy) and the Plextors. But I will not be holding my breath.

Originally posted by Halcyon

2) Pressed C2 errors are only data level errros. Most real life C2 errors are physical level errors. To read these properly the transport, pick up and servo also matter a lot. Even reading the same spot twice might produce unreadable and readable results on the two occasions (this is what EAC uses too).

Actually, from what I have seen, most of the Abex tests are for physical defects. I was more interested in using specific error patterns to test the performance of the error correction and c2 reporting completely isolated from servo/laser/EFM/Synch performance. It should be possible with sufficient tests to determine quite a bit about the strategy and performance of the particular CIRC implementation, its interpolator performance and subsequent C2 reporting. The advantage of using artificial errors is that the results are repeatable and you can have a genuine comparison.

Actually all errors are physical errors. The difference is how those physical defects are formed. There are random errors which can only be one symbol (byte) maximum length or one malformed pit (one of the interleaves seperates the possible two symbol burst that some malformed pits would create). These errors can only occur at the mastering/burning level and because of their size cannot be generated by 'damaging the disc'. The other errors are burst errors as generated by scratches, dust, etc. However it is often useful to look at burst errors as short or long bursts as different CIRCs have variable performance for each case. Most discs contain a combination of both with the random error rates having a considerable impact on CIRC burst error performance.

Originally posted by Halcyon

3) Could we have a STANDARD physical layer error disc (not just data layer) that had scratches of several levels of severity. This would not only test the C2 ability, but also the pickup/servo/transport performance. As such it would be a much better approximation of real life performance with scratched discs.

The Abex and Philips test discs already do this with spots,bars,etc. The trouble is that using large defects to test CIRC and C2 can only give you results for high error rates. In my opinion these error rates are so extreme that they are almost irrelevant for most EAC users. None of these technigues will test a drive for low error rates as is common on most CDs. Just how inaccurate at low error rates are those drives that rate poorly at extreme error rates? Are the Plextors as bad as some tests would indicate or do they just have problems with severely damaged CDs?

Originally posted by Halcyon

PS In a way I hope that we are discussing the remnants of a dying standard which will hopefully be supplanted by SACD/DVD-A or something else... Redbook implementation has way too much black magic associated with it to be a contender for a reliable standard in audio delivery in the 21th century, imho.

There is no 'black magic' in red book. Unfortunately Philips/Sony missed a couple of details that makes audio ripping 'interesting'. Certainly SACD/DVD-A have enormous advantages over red book in booth decoding accuracy and digital audio performance. However both of them are based on DVD disc standards which are more expensive to produce (especially the hybrid discs).

Also there is the issue about the source material, much of it is very poor and most old analog master tapes continue to deteriorate, whereas many early digital transcripts had problems especially with 48ksample recorders which then had to be resampled for CD.

In was my hope that DVD would not have caught on as well as it has as I believe it did not offer enough advantages to be be the replacement for CD, perhaps Blueray ML with its enormous storage density will be a better replacement. Philips have been pushing their 30mm 1gig+ disc and it does have charm but are those cover notes going to be hard to read!

Regards,
Bob

[cdrinfo]

Hi people,

Sorry for not visiting so often EAC's forum but as you may have noticed we are very busy testing many new drives. First of all thanks for your positive comments and your feedback for our tests. I will try to answer, comment some of the previous posts:

Almost ALL manufacturers use ABEX discs for testing/calibrating/fine tuning their drives. Its true that the ABEX discs passes the limits of CD standards and show how a drive will perform in extreme cases.

That doesn't necessary that a drive performed well with an ABEX disc will perform also well with a real life scratched and dusted disc. Our tests with a BAD DAE CD-R media have shown that. The problem is that most manufacturers saying "...we cannot re-produce your results with the BAD DAE CD-R media so they are not useful for us...", and that's partially true.

So here we have the question, test with discs that have repeatable results, or test with unknown (aka hand made) discs? For example, we could create a disc that no drive could pass (remember that there are limits to what a drive can correct), but how this could either benefit manufacturers to fix bugs or show users which drive is the better/worst?

Apart from the test discs, we are using the EAC CD-R test disc. The results with that disc should be partially comparable with other user's results, in case we have done the same defects. As you may have read, many users created test discs that produced a LOT of errors, so their results wouldn’t be comparable with ours/or other users. Unfortunaly, that's the problem here. Repeatability!

To add something useful, CDSpeed has lately added their implementation of C2/DAE Quality results either with the ABEX test discs OR by using the CDSpeed Advanced DAE Test disc (obviously damaged). CDSpeed author said that the patterns used in that disc are not very easy for a drive to correct, so this new test should show some interesting results also.

The CDSpeed and EAC measure differently C2 errors/DAE quality according to each author ideas. We describe what exactly CDSpeed measures at the Writing Quality article.

PS2. A new article that describes the various used test discs is up at: http://www.cdrinfo.com/Sections/Arti...+Media&index=0