Bits are bits

[Cosmik]

If he have access to the master file no doubt.

So in that case why would bit perfect = BS?

Is "access to the master file" the ace up your sleeve, in that it leaves the vast majority of audiophiles (without master files) still wondering whether their bits are defective?

 

[Cosmik]

Some blurb from a CD mastering company:
http://www.cdmasteringservices.com/digitalerrors.htm

WRITE ERRORS
Write errors can result in CD failure. CD manufacturers will reject a master disc if write errors exceed their specifications.
There are three common errors types that affect CD quality. They are C1, C2, and CU.

C1 ERRORS
C1 Errors refer to the block error rate (BLER), which consists of bit errors at the lowest level. C1 errors are always expressed
in errors per second. All CDs and CDRs contain C1 errors. They are a normal result of the write process. However, the
maximum C1 error rate for a quality recording is an average of 220 errors per second based on 10 second samples.

C2 ERRORS
C2 Errors refer to bytes in a frame (24 bytes per frame, 98 frames per block) and is an indication of a CD player's attempt to
use error correction to recover lost data. C2 errors can be serious. In theory, a CD player should correct them. C2 errors
are usually an indication of poor media quality, or the failure of a CD burner to produce a quality burn (see conclusion).

CU ERRORS
CU Errors refer to uncorrectable errors that are present after error correction. No CU errors are allowed in a recorded disc.
Generally, discs with CU errors will not play properly because they contain data that cannot be recovered.

When errors are the result of physical damage to the disc, CIRC Logic - Cross Interleaved Reed-Solomon Code, is used to
identify and correct random errors, which allows some damaged CDs to play without any distinguishable difference.

CONCLUSION
CD replicators consider a disc with an average of 220 C1 errors per second, "a good quality disc." Typically, our masters
average less than 1 C1 error per second with absolutely no C2 or CU errors. We have our own standard which states that in
addition to no C2 or CU errors, we will not ship any disc that averages more than 2 C1 errors per second. That's .009% of the
maximum allowed for a good quality disc. This provides you with an excellent master of the highest quality. If you're going to
have your CD professionally replicated, there is no substitute for a quality master provided by a professional mastering facility.
While mastering is about EQ, dynamics, song levels, etc., it's also about providing you with a low error master that will be
accepted by CD replicators. We test each master for errors before it leaves our facility to insure disc quality.

 

[Cosmik]

You are patronising..,

Says the man who posted a cartoon to ridicule one of my comments.

I'm just trying to get to the bottom of what it is you are claiming. It is not obvious - you prefer cartoons and Chinese proverbs to just stating your case.

 

[Cosmik]

Sorry it wasn't my intention and I apologise if I did hurt your feeling.

No need. I apologise, too.

I think I know what you are saying, but I am not sure whether you are claiming that uncorrected errors are causing audible differences.

 

[Ron Party]

I read with interest the substantive discussion. I have to add that I, for one, enjoy the proverbs as well.

 

[Cosmik]

Can I summarise what I think is being said?

All digital storage media can have low level internal bit errors. Redundancy and clever 'book learning' maths (N.B. Mivera) allow these errors to be corrected so well that uncorrectable errors are rare, making 'IT' practical and reliable.

CD audio as a storage medium is less reliable than many other forms of digital storage, making uncorrectable errors more likely.

A CRC is not a foolproof way of verifying the contents of a digital file or stream but the odds of an erroneous match can be made very low indeed. Weaker forms of CRC are obviously less reliable. A direct comparison between every element of two streams or files is foolproof.

The controversial part: uncorrectable errors in CD are frequent, but because of the ways CRCs are used, we cannot know how frequent..? We don't have access to the master file. Even if we get a CRC match, there may be all sorts of errors going un-noticed. But because CD players can interpolate around errors, they are not audible anyway. The point is that bit perfect hardly ever happens in CD audio.

(not what that CD mastering house claims)

Is this along the right lines?

 

[March Audio]

If he have access to the master file no doubt.

The master file is irrelevant. If I can recover the data repeatedly consistently from the CD (which is the most vulnerable part of the chain) then there is no need to worry about the data further up the chain.

The error correction does its job.

 

[March Audio]

Can I summarise what I think is being said?

The controversial part: uncorrectable errors in CD are frequent, but because of the ways CRCs are used, we cannot know how frequent..? We don't have access to the master file. Even if we get a CRC match, there may be all sorts of errors going un-noticed. But because CD players can interpolate around errors, they are not audible anyway. The point is that bit perfect hardly ever happens in CD audio.

(not what that CD mastering house claims)

Is this along the right lines?

But this train of thought can be validated. If correct, the method I suggested above would contain non zeros because recovered audio would by definition change dependant upon the error.

 

[Brad]

To reiterate, CRC is not part of the reed Solomon algorithm. The parity bits are. My experience is with turbo code and LDPC, but as long as it's the same principle, the CRC argument is a red-herring

 

[Cosmik]

But this train of thought can be validated. If correct the method I suggested above would contain non zeros because recovered audio would by definition change dependant upon the error.

Definitely, but there was the stipulation by alfe that your test would have to applied against the master file. Without that, he thinks that the case hasn't been proved.

 

[March Audio]

Well the technique works. Just loaded a file (from NAS) and mix pasted it to itself in Adobe Audition. Without inversion the levels are additive, with inversion it completely cancels out.

Original

Inverted

Mix paste normal + normal

Mix paste Normal + Inverted

Now to do this with multiple CD rips.

 

[Cosmik]

Now to do this with multiple CD rips.

I guess he would say that you might get lucky with a few CDs and get all the same result, but you wouldn't have proved that the mastering itself didn't contain some errors. Me, I'm more optimistic!

 

[March Audio]

Definitely, but there was the stipulation by alfe that your test would have to applied against the master file. Without that, he thinks that the case hasn't been proved.

Easy way to test that, burn a CD from a track on my NAS. Rip multiple times and compare to original. Place defect on CD, rip again and compare.

We know without doubt that the data stored on PCs (or related server, whatever storage format) is robust. Otherwise we would all be having problems in our daily use of computers. That clearly doesn't happen so I just wont entertain an argument otherwise. There is no reason to question the source data. So the only part of the chain that needs checking is the final CDs because they are possibly more vulnerable to data integrity issues due to physical damage or production quality.

 

[March Audio]

Then from now on you have to buy a bunch of drives to compare and do multiple rip and a null test.

No. I don't suffer from paranoid audio nervosa.

I am quite confident about the error correction capability built into CDs. Also I ripped my CDs with DB Poweramp. This produces a checksum of the track/disc and checks against a database of other users rips of the same CD. It also reports unrecoverable errors. Guess what, no problems found and matches other users.

 

[March Audio]

Then we are back to reliability of checksum.

No, we are back to you trying to make something out of nothing. So thousands of users all come up with the same erroneous checksum. Really? Also note the bit about reporting unrecoverable errors (C2 etc).

 

[March Audio]

So? Seems like an erroneous obsession in that case.

Lets try again. So you think thousands of users worldwide all come up with the same erroneous checksum?

OK below is a bit of marketing blurb but,

AccurateRip

Overtime AccurateRip can become like a wise-friend, someone you can rely on and trust. It works by storing peoples ripping results and comparing your result with theirs. For example 100 people rip Madonnas latest CD, of those 100 twenty have errors, the other 80 all have identical rips. If you were to rip your Madonna CD there are 2 possibilities, AccurateRip would report that 80 other people agree with your rip (confidence of 80), or that 80 disagree if your had errors. What are the odds of 80 people agreeing with your rip, but they really had a bad rip (ie those 80 people had bad rips which happened to give the same check code)? the odds are 4 billion x 4 billion (repeated 80 times), an astronomical number. If more than 3 people agree with your rip, it is 100% certainty it is accurate.

 

[Cosmik]

I'm not obsessed I'm just saying nothing is perfect,otoh I'm an R&D guy I have to
Lets imagine that the majority of the drives in your democratic database are low cost ones with poor performance .
100% certainty is just like bit perfect.( I'm a fan of Heisenberg)

Isn't that a bit 'philosophical'? It isn't 100% certain that the sun will rise tomorrow - nor even that it has risen today. If hard drives routinely had errors, nothing software-based could work. The entire computer-based world must rely on close to zero errors - and it assumes zero errors.

If I wrote a simple program to continuously write patterns into files on a hard drive, then read them back and compare, would you bet a large sum of money on finding an error in an hour, day, week, year? It could probably do millions per second so that would be a lot of potential places for an error. My experience suggests to me that such a system would probably run for years without an error.

 

[Cosmik]

An article on hard drive bit errors. They do occur, obviously - maybe I was (slightly) too optimistic above for a consumer grade drive.
https://www.theregister.co.uk/2015/05/07/flash_banishes_the_spectre_of_the_unrecoverable_data_error/

If I am understanding this correctly, cheap (consumer) drives expect a bit error every 12.5TB. If a normal CD album holds 650 MB then this results in 1 bit error every 19,230 albums.
Expensive (enterprise) drives increase this by a factor of ten. SSD then boosts this by 10, 100, 1000 depending on the grade used.

So of course, never 100% certainty, but reasonable odds of getting a bit perfect album, I would say.

Edit: and I suppose there will be nuances in the stats that come up with the 12.5TB figure - is it a case of an even distribution of bit errors against volume of data written/read, or are there basically 'bad drives' and good ones that, if they could be selected, would behave perfectly for years?

 

[RayDunzl]

At least we don't have to worry much about electron decay. Although, as we approach single-atom storage cells, it might become a consideration.

Presently the estimate for the lifetime of an electron is a minimum of 66,000 yottayears (if wondering, that's a lot).

---

As a humble consumer of consumer audio, considering the other local storage/distribution alternatives:

Vinyl

• click
• pop
• wow
• flutter
• rumble
• warpage
• concentricity
• cleaning
• scratch and dent
• gouge
• track jump
• stuck groove
• stylus replacement
• play time

Tape

• stretch
• dropout
• rewinding
• print though
• pre echo
• hiss
• entanglement
• crease
• head troubles
• complex mechanisms

I'll take an occasional uncorrected music data error that the system hides from me until it gets really really bad, at which time I'll stick the offending disk into a computer drive and let EAC mull it over for a while then burn a new (largely, if not completely) corrected copy (complete with new media errors to correct/hide since that is the nature of consumer optical media).

I've had to do that so many times (once? or was it twice?) I can just barely stand living any more.

 

[RayDunzl]

The only unrecoverable fault I remember having with a CD was one that got caught in the hinge of the jewel case and the case lid manged to clip the edge of the CD right off, like a brake press upon opening.

Surprised me a lot. The last track on that CD is a fail.

It hasn't happened since, but, look out!