Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[David Glasser]

bblackwood wrote on Sun, 11 September 2005 04:25

One of the most important jobs my assistant has is cutting parts - I'd never feel comfortable shipping a master without actually listening to it. The cliche of 'never turn your back on digital' comes into play. Like Dave, we charge for it, and the benefit is we don't ever have to assume anything - we know the master(s) is/are exactly as they should be when they leave here...

In CD world, agree totally. In the case of SACD however (at least using the Sony Authoring tools - I don't know about Sadie or Phillips), AFAIK you can't listen to the final master which is an SACD disc image on AIT tape. Of course we listen to the final DSDIFF files that comprise the master, and then rely on the verification and compare functions of the authoring program to insure that all is OK. At some point, you have to trust your tools (and know when not to trust them). What sayeth you all to this?

[bobkatz]

Ronny wrote on Sun, 11 September 2005 10:53

There can always be a defect in the media itself, Bob.

And that won't transfer to the clone? If there is an error in the source that produces an audible problem, that will be transferred to the copy.

The master disc has been error checked for media errors to be extremely low or it won't pass muster.

Each time a compact disc is played, slightly different correctable errors in different places on the disc occur because there is some "slop" to the design, a little spec of dust here or there, hopefully all of that is correctable. The ONLY time an audible error is generated is if there is an uncorrectable error. Each time you play any disc, its audible output will sound the same UNLESS there is an uncorrectable error, which will result in an interpolation that could be inaudible, or have a tic or pop.

Whether you listen to the original physical disc or whether I listen to a copy of that original disc, there is no guarantee that the next playback will be identical EXCEPT that we count on the fact that the error rates were measured low and that the system is robust. Can you be ABSOLUTELY certain that any particular moment of playback is not a "smoothed out" interpolation? Most uncorrectable errors remain inaudible because the interpolator produces a benign sonic result.

However, since the error correction mechanism on CD is extremely rugged, since uncorrectable errors are very rare on vetted media, and since we had already tested the medium for low errors, the chances of this happening on any play are slim.

But it could happen. So any time ANY of us listens to a disc, perhaps an uncorrectable error might creep in. If we really want to get serious, we should put idiot light error readouts on a CD player to accompany our listening QC just as we had with the 1630. And a timecode readout as to where on the disc that error occurred. But we don't do this, because the error correction mechanism on the CD is extremely robust; we count on that during our QC process.

My point is that if the uncorrectable error does happen, it will happen on the playback of the source disc OR during the dubbing to the new medium (hard disc). I count on this principle when I dub from the playback CD to my hard disc for the second generation copy which generates the QC listen while we cut the CDR QC copy in real time. There is no effective difference as far as I'm concerned.
And the sealed hard disc is a much more reliable mechanism, it doesn't even need a CRC check.

If we hear a problem, we go back to the source files to see if it's there. Same as you would if you heard a problem on your QC. Only we're one step closer than you and even more precise and quick... We're listening on a networked SADiE system in a SADiE project. All we have to do is mark the time code where the questionable noise occurred, WHILE we're making the QC copy. Then we open the source EDL and immediately listen to the source. IF there's a problem that we should have caught, it goes back to the mastering studio, where I denoise the click or tic or fix the problem, and then rerun that portion of the master through our processors, which then gets edited back into the capture file and a new master is cut with the correction.

10 years ago I heard an engineer rave how great a particular new CD blank "sounded" because it had the lowest error rates he had ever measured. Give me a break! This is suggestability at its height.

Years ago we discovered that as 16-bit PCM-F1 tapes aged, little tiny "tics" would start to become audible in soft passages. This was an indication of the beginning of uncorrectable errors. To make masters from these tapes we had to dub them several times as usually the tics appeared in different places so we could edit the good parts together. The re-edited copy became the new master. Today we would probably use a declicker to fix the tics.

Every archival organization storing digital media have to "renew" the media, typically every 5 to 10 years. Copies are made and the data is migrated to a new medium. The old medium is, I think destroyed. The entire world relies on this concept----that digital copies are as good as the originals, and from the point of view of error rate, a new digital copy is probably BETTER than an old original.

The only reasonable argument I've heard from you guys in this thread in favor of auditioning the original media is Brad's very reasonable adage "never turn your back on digital". Or "you never know what can happen". You're increasing the chances of error by making life more complicated.

But not necessarily; if we can cut the QC copy that the client is going to audition while we are QC'ing the master, then we are doing the client a service as well. Think about it. We get additional controls over the master, including having an exact clone of the master on our hard disc as a reference, which generates all the QC copies.

And based on my arguments in this thread, it should be clear to you that I feel that ANY error or problem that would occur in the playback of the master will show up on the QC copy. Hundreds and hundreds of approved masters later, I can attest that this is a fact. In the end, regardless of how you feel about my practice, it is true in both cases that only the ears of the QC person separate mastered success from
failure.

BK

[rondr]

Hi All,

Please correct me if I'm off topic here.  How about a test pressing?

Best,
Ron Rice

[bobkatz]

rondr wrote on Sun, 11 September 2005 14:56

Hi All, Please correct me if I'm off topic here.  How about a test pressing? Best, Ron Rice

Most of the plants I deal with will not do a test pressing for typical quantities of under 10,000. I'm sure that Britney Spears gets test pressings, though.

Speaking of test pressings, remember the days of vinyl, when you could not play the lacquer?  Tracking down the cause of tics and pops was quite an art. Mother, father, where art thou?

[bobkatz]

Off topic here, it seems that most of you guys are regarding my position on this topic to be crazy/radical, even "unsafe", whereas on certain other topics I would be labelled a conservative.

No one would deny that I haven't thought this method through very carefully (and conservatively) and that it is methodical, however radical it may seem to you.

As Dave Glasser said, "at some point you have to trust your tools." Hell, I never turn my back on a CDR either... I never assume there is audio on a CDR unless I play it, and I'm certainly aware that you NEVER know if the CDR is good unless you have played it through from beginning to end. You are ALWAYS one generation behind in your evaluation.

As Ronny said, you are listening in one continuous pass to the data that is going to become the master. How many of you listen to the OUTPUT of the dubbing medium in E-E while making that dub? The principle of listening through the last piece of valid electronics in the chain applies here. I'm just extending that to the making of the QC copy while proofing the master.

BK (shaking up some more tailfeathers)

[dcollins]

bobkatz wrote on Sun, 11 September 2005 06:00

Each of you has only offered the ad hominem argument that "you have to listen to the part because you have to?" without providing any evidence that there is anything wrong with the other method.

As usual, I'm just trying to understand why your approach is so, uh, unusual......

Listening to what you are selling seems pretty obvious to me, but why not turn it into some kind of convoluted system with multiple levels and needless complexity?

Quote:

DC digresses into John Vestman arguments which are completely irrelevant to this discussion.

Well guilty on that one.  But only at the end.  What about my other points?

Quote:

Now, What about DDP tapes? HOW MANY OF YOU ACTUALLY LISTENED TO THE PHYSICAL DDP TAPE? RAISE YOUR HANDS . I wager that over 95% of you made CDR copies of it and used that to evaluate the master. Did you feel inadequate, dirty, unsatisfied that you couldn't listen to the part?

DDP tape is dead as Elvis, but had off-tape ECC, so you were checking as you go....

Quote:

What about  FTPs where there is no physical part? (Where does the liability start, and end?)

Much like CDR transfers, I have never had an error in FTP.

Quote:

That hard disc copy is not the physical master. The physical master is the CDROM that will be sent to the plant. But do we "audition" that CD ROM? Can we audition it?

Hey, now we're getting somewhere!

Quote:

Likewise, is it illegitimate to  listen to an audio CD clone of an audio CD? Is that somehow "different" than the DDP (or FTP) scenario I just described?

"Without Benefit of Clergy" copies?

Quote:

At the plant, do they listen to the grandfather, the father, the mother, the stamper, or the pressing? Is there any point in "listening" to any of those stages?

As it turns out, they "listen" to your CDR as it goes on the server (we hope).

Presumably not for the first time........


DC

[Bob Olhsson]

I had a 3 second dropout happen in the middle of a track on an Exabyte DDP tape and learned the hard way that nobody at one of the most reliable plants listens all the way through.

Now that CD replication has become a generic commodity, I'm afraid the days of audio QC at replication plants is long gone.

[bobkatz]

dcollins wrote on Mon, 12 September 2005 00:02

As usual, I'm just trying to understand why your approach is so, uh, unusual...... Well guilty on that one.  But only at the end.  What about my other points?

Well, it works, for me. And your points are perfectly valid, and I have listened. But it doesn't deny the validity of the fact that any defect on the source will show up on the copy. So I see no reason to stop doing what works.

You said, it is "more complex". If you had a SADiE system you would see that CDR loadback and DDP loadback is totally integrated with the system; you don't even have to think twice. It is not a complex operation. Now please don't anyone go accusing SADiE of having encouraged this method.  I'm the one who's doing it.

It seems extremely radical to you, and maybe they'll kick me out of the union for saying this, but I maintain that listening to and approving the Japanese master which is a copy of the European master which is a copy of the American master EXACTLY (not "effectively"----EXACTLY) verifies all the previous masters. Providing you did a BLER test of each of the physical media.

Do your clients listen to the physical master? Most clients make their judgments from the copy.

It is fair to say that I have passionately listened to all of your passioned please and that I appreciate your concerns! I understand that Glenn Meadows' used to have his QC person do the QC while making a copy. I think they took the SPDIF output from the Clover. Sounds less reliable to me than a CDR copy in the computer.

Anyway, if his QC person listened to the output of the digital copy machine while making the copy, does that make his listen invalid? What if he waited to play back the copy after the copy was made?

If we hear a noise or a problem on the copy, we'll find it on the master! So far, in hundreds of masters, we have NEVER found an audible problem on a copy that cannot be found on the source. It is possible that an uncorrectable error will occur on a copy that cannot be found on the source----however logic makes it very clear that any problem that's found on a source will make it to the copy.

At least that's how I say it.

Quote:

Listening to what you are selling seems pretty obvious to me.

Fair enough. DC, you have the last word on this. I'm a minority (of one?) that has many satisfied clients who know that when Digital Domain approves the master, they approved the master.

Quote:

DDP tape is dead as Elvis, but had off-tape ECC, so you were checking as you go....

ONLY THE MEDIUM, NOT THE SOUND. You can have a DDP tape that's totally blank but meets all the ECC tests.

I recall an instance when someone cut a master on a CDW-900E and it error tested fine. Shame that they didn't listen to it, because there was a memory chip failure in the 900E and no sound got on the master at all. "Gee, it sounded fine going down." Both you and I would have caught that with our current verification methods.

Quote:

Quote: At the plant, do they listen to the grandfather, the father, the mother, the stamper, or the pressing? Is there any point in "listening" to any of those stages? As it turns out, they "listen" to your CDR as it goes on the server (we hope). Presumably not for the first time........

Hey, why not fix it in the shrinkwrap?

We can only hope, but I believe that the Eclipse software just makes a high speed digital to digital transfer from the CDR to a DDP image file onto hard disc. It's all automated as the plant personnel are just transfer monkeys. If there is a listen, I suspect it occurred somewhere in the premastering department.

I suspect there is no error checking done during the transfer from our CDR, unless Eclipse has some kind of proprietary CD ROM drive that delivers an error readout while the copy is being made. That would be very encouraging.

Anyone from a plant can confirm or deny that suspicion?

[Jerry Tubb]

This discussion should include yet another facet:

We're doing a fair amount of 5.1 surround and 2.0 stereo mastering for DVD video,  Recently added DTS and AC-3 encoding, and send the soundfiles on hard drive and/or data DVD-R to the Authoring house.

We Audibly proof the audio before it goes to Authoring. The Producer usually attends the Authoring sessions for QC. DLT tape (iirc) goes from Authoring to Replication.

We get a final chance to QC a "test pressing" ref disc before the full run of replication. The whole process introduces a whole new level of chances for QC problems to occur. So we have to rely on communication between all parties involved in the chain, and trust everyone to QC at every level.

[dcollins]

bobkatz wrote on Mon, 12 September 2005 06:02

dcollins wrote on Mon, 12 September 2005 00:02 Listening to what you are selling seems pretty obvious to me. Fair enough. DC, you have the last word on this. I'm a minority (of one?) that has many satisfied clients who know that when Digital Domain approves the master, they approved the master.

It's not about having the last word.  

I just want folks to understand that basic QC should involve the duplication master, and not a copy.  

That's all I'm saying....

Quote:

I recall an instance when someone cut a master on a CDW-900E and it error tested fine. Shame that they didn't listen to it, because there was a memory chip failure in the 900E and no sound got on the master at all. "Gee, it sounded fine going down." Both you and I would have caught that with our current verification methods.

Are you making my point?

Quote:

I suspect there is no error checking done during the transfer from our CDR, unless Eclipse has some kind of proprietary CD ROM drive that delivers an error readout while the copy is being made.

Bob, if you were in the business of CD manufacturing, would you allow it to go on without "error checking?"

I'm not sure what else I can add here.

Perhaps a funny joke..............

DC

[bblackwood]

bobkatz wrote on Mon, 12 September 2005 08:02

It seems extremely radical to you, and maybe they'll kick me out of the union for saying this, but I maintain that listening to and approving the Japanese master which is a copy of the European master which is a copy of the American master EXACTLY (not "effectively"----EXACTLY) verifies all the previous masters.

This kinda throws me, Bob. Are you saying that once you cut masters (for multiple releases) you make copies of it, 'daisy chaining' the copy process, then finally make a ref copy of the last one to QC?

[Thomas W. Bethel]

Ronny wrote on Sun, 11 September 2005 10:53

There can always be a defect in the media itself, Bob. We all hear the whole song when it is a file as we process. That's where we listen to original data that is to be cloned. IMHO, it's still beneficial to audition the "media" that is going to the plant. FTP, has a check and blance function that is very simple, if any data is corrupted on the transfer, the checksum won't match and you get notified. Although I've experienced this on rare occasion, it was probably due to the transfer itself and just redoing the transfer has always corrected it.

With Wavelab I can do a comparison of what I have on the computer versus what is on the master disk (after loadback) and it will tell me where they differ. I can do this for every master we send out but most of my clients don't want to pay for the extra time involved. In fact most of my clients don't want to pay for any QC'ing done by us and prefer to do their own listening and auditioning.

I had a batch of Kodak disks (back when Kodak made CDRs) that for some reason were not 100% reliable so we auditioned everyone before it left here for the plant "just to make sure" If we had used Bob's method we could have sent out a bad master without knowing about it since we did not audition the master CD. I wish I could do that today but with our current staffing and current work load it is just not possible unless the client wants to pay us for the work.

I believe that the client, if they don't want to pay us to do it, HAS to be the final QC and if there are problems I expect them to get back to us ASAP so the problem can be addressed before the CD is sent to the plant. If this means that have to listen to the master CD from beginning to end then that's what they have to do.


MTCW

[TotalSonic]

bobkatz wrote on Mon, 12 September 2005 14:02

I believe that the Eclipse software just makes a high speed digital to digital transfer from the CDR to a DDP image file onto hard disc.

It's possible now with Eclipse to stream directly from the optical read of the CD-R to the LBR.  Some plants transfer first to hard drive or Exabyte though.

Quote:

It's all automated as the plant personnel are just transfer monkeys.

While most people working in Europadisk's glass mastering dept were more straight clerking techs that might qualify as "transfer monkeys" there were a couple (the department head and one of the senior techs) that had their Electrical Engineering degrees, were mechanics able to fix nearly every piece of equipment in the plant, and knew a heckuva lot more about the details of optical disc technology than nearly anyone that posts on this board.

Quote:

If there is a listen, I suspect it occurred somewhere in the premastering department.

At Europadisk, unless there was a problem during glass mastering or replication, the only listening to the audio master at all was done by the customer service reps when the job was initially received.  A few seconds of each track would be listened to to verify that the music wasn't bootlegged.  Sometimes even this step would get passed by if the CSR's were slacking.  I have a feeling most other plants do the same thing.

The expectation that a plant would devote the enormous amount of manpower to listening through an audio master all the way through while they have to process hundreds of jobs a day is completely ridiculous to me - especially considering how tiny the profit margins are on manufacturing CDs these days.

While working there I was once in a while asked to evaluate replicated product vs. the original master if there was some potential issue being QC'd for - but this certainly wasn't s.o.p.

Quote:

I suspect there is no error checking done during the transfer from our CDR, unless Eclipse has some kind of proprietary CD ROM drive that delivers an error readout while the copy is being made. That would be very encouraging. Anyone from a plant can confirm or deny that suspicion?

Eclipse is actually a suite of applications.  One of them is a tester which gives an extensively detailed report on BLER, CU's/E32's, TOC & subcodes, conformance to various book specs, etc., along with a "Pass", "Warning" or "Fail" grade to the CD-R master.  The testing involves reading the disc image stream at the same speed it would be transferred to the LBR at (usually 4x).  The CD-ROM drives used at Europadisk were non-proprietary slightly older Plextors.   A single CU would generate a "fail" grade - and a cluster or higher level of C2's would also generate a "fail".  Occasionally some fault in the TOC or subcodes would also generate a "fail"

So  - an initial error checking on the master was always done. However - if the problem on the master is actually audio and not an error - it would not get picked up by the plant.

Best regards,
Steve Berson

[TotalSonic]

David Glasser wrote on Sat, 10 September 2005 17:15

bobkatz wrote on Sat, 10 September 2005 09:46 The "fingerprint" thing is strictly related to who pays for a replacement master if there is a failure at the plant. My exerience, for the handful of times it's occurred, is that the plant will NEVER agree to pay for a replacement master, even if the 'problem' is clearly on their end, and even if the master was accompanied by a QC report. This happened once with an SACD master on AIT (that in retrospect was A-OK), and those masters are costly compared to a CD-R. I asked said plant if they had looked at the logs and QC report supplied with the master and was told that they NEVER LOOK AT THE PAPERWORK. EVER.

Can't speak for all the plants - but at Europadisk the following info WAS looked at on the paperwork:  total length of the program, total number of tracks, type of master (i.e. CD Audio, CD-ROM, multisession, etc), presence of CD Text.  This info was recorded in a database and used to verify that the job was being correctly replicated.

Quote:

In their view, the onus is always on the customer. This is a sorry state of affairs. I'd be interested in hearing others experiences with pressing plants.

My sympathies are with the pressing plant on this one.  A provided QC report can say whatever the provider wants it to and is essentially meaningless as far as liabilities go.  The master received often takes some detours before it gets to the plant - i.e. sometimes the label turns in refs or dupes instead of the factory sealed  copy.  And while the ME's cost for the master is in time invested and the fifty cents for the piece of plastic - the plant's materials costs can get to the thousands of dollars.

Best regards,
Steve Berson

[TotalSonic]

I think this discussion has been good for me in that it made decide that since so many indie artists are always looking for ways to cut corners with costs that when I open my room this winter I'm going to have two different prices - one lower one for a CD-R "reference" master - and another for a CD-R "sealed replication" master which would include the cost for the studio time for complete audio audition.

Best regards,
Steve Berson