Guys, thank you so much. You both make great points. But absolutely I'm doing a DDP the next time I do any audio production for a large CD run because it just makes more sense to have a guarantee that the file will be correctly delivered - it just makes no sense not to, when you spend weeks creating a project, doing the recording and artwork and setting up the finances, all to rely on a weak link in the chain. Dithering - good points. To date I've never been that concerned because of the nature of the stuff I've been doing - because the noise was well above -96 by the time everything was thrown into the mix. Though that's ignorance talking, and obviously that's something I need to understand more. On my solo harp CD I was very careful about everything I could do to make it sound good - recording in the dead of night, using very finely-crafted convolution reverbs, etc. The final CD sounds incredibly clean and therefore you notice absolutely everything going on (good and bad!) in the recording. However, as I've started to read up on this, particularly its relevance to downsampling, I'm starting to get what you say, @DrDave about dithering. Since you mentioned iZotope, I looked and found this, which is really nicely written (rare in audio!) and which I link to here in case others find it useful: http://www.izotope.com/products/audio/ozone/OzoneDitheringGuide.pdf

Thanks! The only one I know well is their noise reduction, which is really good. Looking forward to having a play with the others - and experimenting with dither!

A lot of good information on this post about CD replication and the digital errors that we come across in the field. A few years ago I used to work for a CD replicator in New York as a pre-mastering technician. It was our job to write DDP masters along with "100%'ing" or monitoring the audio before it enters the mastering stage (otherwise known as "creating glass" for the replicators).

I'm not sure I can add to this conversation since a lot of good facts have already been mentioned by @DrDave and @Mark_the_Harp, but I can surely say that if your replicator is simply "burning" discs from the master, you're getting burned (no pun intended). It's not a professional replicator. Then again who knows how many exist nowadays since the world has turned to digital/MP3 format about a decade ago.

@DrDave In response to your post 3 above this one, you make a good case for having a (set of) guaranteed identical physical discs. And in terms of my original post, your sentence "But if the pressing plant uses the right machine to rip the CD, it will have to "pass" meaning there are zero problems with the disk." is very useful, because I think I can argue that even if there were errors on the master disc I sent them, if they were uncorrectable then they should have spotted that and either used the second copy, or let me know.

I like the "print-on-demand" approach very much. It's similar to the technology that's been used to print foreign newspapers for years now, in mainland Europe - in other words, if you want a UK / US edition of some paper or other, you can get one. It's identical to the one you'd buy in New York or London or wherever, but printed locally, so there are no long-haul transport costs, but more importantly, they can be provided on the same day. The one thing you might miss doing that, is quality control.

You could argue that the answer to CD replication is not to use physical media at all, but for certain things it's good. It's great to have beautiful printed books (essential for our particular project with children as the booklets are handy for the words and for ideas on how to use the music tracks) and people (still) ask for CDs. At gigs people like to buy music CDs because it's a souvenir for them - and you can autograph it. Try that with an mp3!

@Mark_the_Harp glad to hear your intermediary had a back-up, I hope the next batch goes smoothly.

@DrDave if you're duplicating a small order for a quick turn-around, then burning directly from the master (on a small duplicating bay) is typically your course-of-action in pre-mastering. It definitely increases your chances for what happened to Mark_the_harp where certain discs may contain digital errors, but this should have been quality checked by QA on machines that do high speed scans of the discs (typically a few are scanned in each batch or roughly 10%).

On the other hand if you're doing a high capacity order with multiple color print on the discs, professional inlays, etc, etc...then going through both mastering departments is necessary in order for a timely turn-around. The audio is monitored AFTER it is written to the media for the mastering department and this ensures that the audio is not being altered at that stage. As for the mastering and replication departments, they do the aforementioned audio monitoring, error scans and so forth to ensure reliability. No sound is being altered in any given way since the media is literally a digital duplicate of the master. Pre-mastering and mastering departments do not alter audio unless the client specifically asks for a change. Where I worked this had occurred, albeit rarely. Typically the audio engineer had forgotten a step, made a bad transition or edit, didn't EQ the way the artist intended or whatever may be the case.

@DrDave thanks man! Sometimes I wish that I still worked in the CD replication industry, but that was well over a decade ago. Nowadays there's very few professional replication houses around due to the MP3/iTunes transition. I guess like anything else in the technologically sector, accepting change is part of the process.

Back in the day any CD's that failed their error scans were called "coasters" since their only foreseeable use was as a drink coaster. I'm guessing that applies in your case and I'm sorry to hear it. Anyways I understand your frustration - if there's anything I can do to help let me know.

You have a common problem. I have heard worse, seen worse, and have two examples of major record labels where they actually used audio from another project, so we had to go through thousands of CDs, open them, check that they had the right group playing, then reseal them. I have seen several times runs of 2,000 that had to be dumped. I have seen boxes of discs that were unplayable. I have seen big dropouts, so that there is a loud click in the audio. Basically, absolve yourself of the idea that these places are like a doctor's office, where highly trained professionals are examining every bump on the CD with a scanning microscope.

The fact is, that there ARE some highly trained pros working in some of these places. But there are also some some seriously hung over rockers :) If you had Killagram working on your project, you would not have this problem.

One place I worked with, they printed the CDs, and they didn't like the color, so they redid the whole run. I never saw the ones where the color was off, they just dumped them and redid them, and ate the cost. So guess what, they always get my business.

A lot of these discs, no one ever listens to them, or if they don't play, they just toss them. So you don't hear about it. Suppose in your run ten percent were bad, and you sampled one or two. You probably would never know.

I would say if I look at the last 100 projects, five had to be trashed, and another three or four had serious problems. So under ten percent, but still a big problem. If I go back to the previous 100 projects, the mastering was better and the CDs were better, but still a few of those had to be trashed as well. The real problem is if you have a thousand CDs and there are, scattered in there, 100 or so that are bad. There's no way to know unless you go through each one. So I usually take five-ten random samples, and put those on the Plextor to check them. I'm going to give away at least that many to radio, reviewers, etc, and that way I know that the review copy is perfect, so I solve two problems at the same time. Last thing you want is the reviewer to call you up and say, hey, it won't play in my car stereo (yes, this has happened).

I also built a duplicating tower from Plextors ( I now have a Blu-Ray tower that also does CDs pretty well). You get an "Acard" and some burners, and you can make perfect CDs. Using specially surfaced CDs, like TY "watershield" and say an Epson Artisan printer, you can make awesome CDs. You can even master HDCDs (you may need a license), which really do sound better. And you can do lightscribe in a duplicating tower, so that when the CD is burned, the tower will burn the artwork right into the other side of the CD (you have to flip the CD, DOH!) But who has time? Looking ahead a few years, it could be that the personal tower is the way to go, because physical CDs will be a tiny portion of the market, and you can then make audiophile CDs for those who want them. You can even get a special Yamaha burner to make CDs with larger pits that are a notch better than the Glass Mastered ones.

When you release the CD, release some video as well to market it. You can even burn a short Blu-Ray onto a regular DVD and have that duplicated en masse as well, then put the two into a double case.

If I have some time I will post some scans from some glass mastered and burner farm runs--you won't believe some of the error rates, and there are some perfect CDs as well.

Thank you both SO much - I was getting so paranoid that the plant was going to blame me for this somehow, and it's really depressing when you've done a ton of artwork and recording to have this happen. I have to smile at your description @DrDave as that's exactly how I imagined a CD pressing plant to be - all protective clothing, white gloves and hermetically sealed rooms where scientists sit there carefully polishing the pits in the masters before lovingly and delicately loading them onto the machines.

Hi again - while I'm waiting to hear about this, can I ask a really dumb question: So my master got sent to the plant and failed the error-checking process at the plant. The intermediary company had my second master (the backup) but didn't send them that (I assumed they sent both masters on to the plant, but no, they keep one for themselves). So the intermediary made a copy of my backup master and sent that to the plant. The plant got this copy of the master and used that to create their glass master and produced the really horrible stamped CDs without presumably ever checking that it sounded really horrible.

I'm really puzzled about something: OK I didn't make a third copy of the master that I kept myself, but for all I know I can only assume that my backup master also had errors. I don't understand how the second disc (the copy of my master that the intermediary sent) didn't fail the error-checking process. In other words, if the intermediary company made a perfect copy of my backup master, wouldn't the copy that they sent over have errors too? Or would that depend on how they made the copy of my master? And, come to that, wouldn't a bit-for-bit digital copy have had the CD-Text that was on the master but somehow failed to get onto the pressed CDs?

And basically, wasn't that a really odd thing to do, when the first master failed the error-checking, to send a copy of my backup master rather than the master itself? I'm thinking that this whole process (if you can call it that) is what caused the pressing company to think they had an error-free CD and simply going with it, regardless of the fact it sounded horrible.

Hope that makes sense - it's a bit complicated to explain (diagrams would be easier!!)

@Meierhans Thank you - yes, you're absolutely right. I have a very cynical view of businesses, but the fact they left the CD Text off is on its own a cause for having the work redone.

I could not believe it, when I found out two years ago, that CD-manufacturing companies accept burned CDs as masters for the printing process. The worst medium for storage and transfer. As long as defects do not exceed a certain size (I do not remember percentage, but the physical gap can be around 1.4mm) parity bits help to reproduce the original data without loss and without the need of interpolation. But burners can malfunction, disks can be burned too fast or have already aged too much, because UV light and oxygen attaced the dye because of poor manufacturing. Modern error interpolation will mask such errors, until they are litterally printed from a defective master.

I sometimes realign my CD-Players by monitoring the lasers´ eyepattern. Old machines, one already 28 years old, but completely serviceable and adjusteable. Printed CD´s have a clear, sharp eyepattern, but with burned CDs, even fresh ones, the eyepattern will show more noise, because the edges of burned pits and lands are blurred.

Hi both - the intermediary admitted it was their fault or (to put it in their words, "we think this is where the issue probably occurred"). Basically I'm pretty sure both my master CDs had errors on them, which is obviously bad on my part and stupid of me not to check - the fact the second wasn't picked up by the plant suggests yes, that the intermediary should have sent on my second CD, then they would have found the errors on that and I would have simply sent better masters.

I believe @killagram that what you suggest is exactly what happened - that the intermediary simply copied my CD without checking. I can't imagine why they didn't just send my original backup master, as the replication plant would have picked up the errors on that, just like they did with my first. And yes, a huge pain in the ass, not to mention the worry I've had that they would try to wriggle out of it somehow.

Anyway I insisted they collect all the CDs which they've now done. I'm guessing they don't handle DDP files as they were copying masters and sending them on to the plant - so I didn't go down that route. I've burned two new masters and surface-checked them for errors, and also checked with EAC using wav compare that the wavs extracted from the first production run done a few years ago (which was perfect) are identical to wavs extracted from the masters I've just newly created for replication. They are identical. Of course, I can't simply send them that CD from the previous run as it never had CD text as I hadn't bothered creating it for that original run.

So...a happy ending (I hope - I'll know when the CDs come back!) but also a good lesson in how things can go wrong sometimes.

Hi @Maria2345 not sure what you mean by

The defect may be on the sound element in the sound box

...but I got my re-pressed set of CDs back and they're fine. Well - when I say fine, they're still missing the CD text. So lesson for me - use a DDP next time.