Review: EMI's Copy Control

A friend loaned me a copy of Seven Circles by The Tea Party.  This is a disc of the kind you've probably heard of; it superficially resembles a CD, but technically is not one.  It seems to be published by EMI. My friend was unable to listen to it on equipment designed for playing CDs, and asked me to examine it.

At first glance, it sure looks like a CD. You have to look pretty hard to notice that the "CD Digital Audio" logo simply isn't present anywhere on the disc or its packaging, which otherwise resemble those of a CD. There is a box of text on the back advising that "On some equipment, for example car CD players [comma missing in the original] playback problems may be encountered" but also claiming that it's compatible with all CD/DVD-V/DVD-A/SACD players, PCs with Windows 95, a Pentium 2 233MHz, and 64MB of RAM, or a Mac with MacOS 8.6-9.2 and CarbonLib, or MacOS X.

The only CD players I have are attached to my computers.  The ones I tried the disc on are a Yamaha 8424 external SCSI CD burner and a Toshiba internal IDE DVD-ROM drive.  The Toshiba (driven with workbone under Linux 2.6) plays the disc as a normal audio disc, except that it can't seek from the last track back to the first track; workbone gives a message about the drive reporting an error when it tries to do that.  I didn't try to play the disc with the Yamaha because the speakers are only attached to the Toshiba.

The Toshiba, driven with cdparanoia on all the default settings, rips the audio from the disc just the same as it would from a real CD. The only visible difference is that cdparanoia returns a lot of correctable errors.  In the status/progress bar (which should show all spaces on a perfect read, but sometimes shows two or three plusses on a dirty disc) it shows about one third spaces, two thirds plusses, and an average of about one "e" per track.  Plusses indicate "unreported loss of streaming", which seem to be situations where the drive can't read the data right on the first try and has to try again, so there's a bump in the flow of data between the drive and the computer.  I've never seen "e" on the status/progress bar before, but it appears to mean an error reported by the drive to the computer but corrected on retry.

At a guess, what's going on here is that there may have been a lot of deliberate low-level errors introduced into the disc (and/or it may be mastered slightly outside the envelope in terms of parameters like pit size or contrast ratio); the idea being that an audio player will probably be able to correct the errors, usually, and maybe a computer player will have trouble.  Adding deliberate errors like that is not really a good thing to do, because it cuts into the margin available in the error correction scheme.

The mathematical code used for recording CDs is designed to be able to correct a certain number of errors per disc block; as long as there are fewer than that number of errors, you get perfectly reproduced music, but when there are too many errors, it tends to fail rather suddenly.  The way the system is supposed to work is that the disc is recorded (to the limits of the technology) without any errors at all, and then as the disc accumulates dust and scratches, those generate errors which are corrected by the math.

It looks like with this disc, there are a bunch of errors built right into the disc right from the start, enough to push it close to the sudden-failure line.  Then if it acquires even a few additional errors from dust and scratches, it'll hit the limit and fail.  If that's what's going on with this disc, then that means the disc will be a lot more fragile than a CD is supposed to be.  Of course, if the disc does stop working as a result of dust and scratches, that may be bad for the person who paid for it, but it isn't so bad for the record company - they get a chance to sell a fresh copy.  Isn't science wonderful?

Note, before I get into further details of the copy protection scheme:  I was able to read the disc with the same software and settings I would have used to copy a real CD. The copy protection would be no inconvenience, and barely noticed, if my aim were simply to copy the disc.  If I did copy the disc in the most obvious way, the copy I made would have all copy protection removed, allowing further copying, and (being recorded without deliberate errors) the copy would sound at least as good as, possibly better than, the original commercial disc.  As far as succeeding in its claimed goal of preventing copying, this copy protection gets an F.

However, on examination, the copy protection wasn't completely invisible.  There were perceptible differences between this disc and a real CD. The first one was the time to correct those correctible errors.  Because of the retries, ripping took a little more computer time (maybe 50%) than would be needed to rip a normal CD. But I don't think that can be counted as any kind of success of the copy protection.  It didn't impose on my own time to do the rip; even with a normal CD, I'd have walked away from the computer and let it do its thing until done anyway.  In actual fact, the copy protection did slow me down in real terms just because I wanted to figure out exactly what they'd done, and I spent a fair bit of time examining the disc with different tools to see what was going on.  Here's what I found.

First of all, it's a multisession disc.  That means instead of just being one groove that starts in the centre and spirals out to the edge, there are two grooves, one starting just a little outside the end of the other.  The possibility for multisession discs exists because of the desire (back in the day when recordable discs were expensive) to be able to record some stuff on a disc, then come back later and record more on the same disc.  It was at that time impossible to pick up a groove that had already been recorded (for mechanical alignment reasons) so drives were designed to be able to search for a new groove starting a little bit outside the end of the previous groove.  Using two sessions is already a little bit of copy protection right there, because most audio players will give preference to the first session and most computer drives will give preference to the second session.

So if you put this disc in an audio player, it will be more likely to see the first session, which contains the actual music; a computer drive will be more likely to see the second session, which contains no music but computer stuff instead.  Of course, there's also some possibility that the audio player will see the second session (which is not audio) and be unable to play the disc, and it's virtually certain now that computer drives are as sophisticated as they are, that the computer drive will either see the audio by default or be easily configurable to see the audio.  So this part of the scheme doesn't work.

The next line of anti-copying technology seems to be in the computer data recorded in that second session.  It's a standard ISO-9660 session containing an "autorun.inf" file pointing at a "player" program.  So as soon as you put this disc in a drive on a Windows machine, the software on the disc will start giving your computer instructions to follow.  I didn't spend much time looking at this part but it seems to be pretty much what someone might expect - the software automatically installs itself on your drive, contains a "player" program giving you some kind of limited access to the tracks on the disc, and apparently contains some kind of measures to make it more difficult for you to copy the disc.  I don't know what measures those are.  They could in theory be pretty nasty; for instance, the software could go looking through your drive for files it considers unacceptable, and delete, trash, or report to a central authority those files.  I don't know for sure that the player software does anything really bad; I did not run it, or seriously reverse engineer it.

Interestingly, I couldn't find any kind of license agreement for the software either on the disc or its packaging.  Maybe if I ran the software on a Windows machine, it might pop up a "Do you agree?" box, but given that I didn't run the software, and I went looking for a license agreement and couldn't find one, I think the makers of the disc would be unable to bind me to any license agreement.  I also think that depending on just what kind of notices the software gives the user, they may be on the line as far as unauthorized use of computer resources.  If this software installs itself (as it apparently does, because there's an uninstall program with a very terse README explaining that the software installs automatically when the disc is first played), and if it installs itself without asking for permission or giving the user an option of NOT installing, and it prevents your computer from doing things you tell it to do, then the player software looks not much different from the average email worm.

This part of the scheme, because it depends on an autorun.inf file and can be disabled by pressing the shift key, or by running an operating system that doesn't do autorun, clearly doesn't work.

Now, looking at the audio session itself, I notice that although it has a basically valid table of contents, all the audio tracks are marked as "data".  They seem to be marked as audio in the actual subcode data on the tracks, so I guess the assumption is that an audio player will trust the subcode over the table of contents, while a computer player might trust the table of contents and say "I can't rip this as audio, it's data".  Note that cdparanoia didn't seem to notice this level of protection, so either cdparanoia doesn't look at that data, or doesn't care.  The copy control fails again.

However, somewhat to my disappointment, my Yamaha drive did have trouble reading the audio tracks.  I'm not sure whether the trouble came from the mismatch between the table of contents and the actual data, or from invalid data in the tracks, because it looks like there may have also been occasional blocks marked as "data" instead of "audio" mixed in among the audio blocks within the tracks.  The Yamaha drive has a very good reputation for being able to read and write anything, so I wonder if the issue may have been in my SCSI driver or in cdparanoia's support of the Yamaha's features, rather than in the drive itself.  At a glance, it looks like cdparanoia was aborting when the Yamaha drive gave errors, where it would have just retried and eventually ignored the same errors had they come from the Toshiba drive.  Better support of the drive in cdparanoia might fix that.

The SCMS bits in the table of contents were all set to indicate "no copies allowed".  Well, as far as I can tell, only some DAT recorders actually pay any attention to those bits anymore anyway, so that has very little effect.  The "pre-emphasis" bits were turned on, too, which is a little unusual.  Nobody actually uses pre-emphasis on CDs, and by listening to the tracks, I wasn't able to determine whether they really were recorded with pre-emphasis, or just marked that way to cause further confusion.  When I made a copy for personal use (as is my right under section 80 of the Copyright Act) I turned on the pre-emphasis bits on the copy.  I can't tell whether it makes a difference to the way the disc sounds.  I'm guessing the pre-emphasis thing is meaningless, because for sure if there were pre-emphasis present to the extent that it's present on vinyl records, I'd be able to recognize that by listening to the data.

The copy, of course, plays perfectly in all drives I've tried it in.  I was actually kind of tempted to clone the disc - dual sessions, screwed up table of contents, auto-running trojan horse player software, and all - but since the point was to be able to listen to the music as easily as possible, that wouldn't really be worth doing except as an intellectual exercise.  Being able to produce such a bizarre non-compliant disc might be an interesting challenge for operators and authors of disc-mastering software.

One final annoyance from the copy protection was that I couldn't trivially extract an ISO image of the data side of the disc.  Running dd on /dev/hdb just gave me an error, even though I could mount the disc fine by specifying the "session=" mount option, and then (if for some bizarre reason I wanted a copy of EMI's DRM trojan) copy the files off as I might copy files from any CD-ROM. I suspect that this issue is simply because it's a multi-session disc, and I could probably get /dev/hdb to cough up an ISO image by reading the manual and maybe making an appropriate ioctl call or something before extracting the image.  It's not worth bothering, though.  Why would I want a copy of a virus that doesn't even work anyway?

So, in summary, the disc has a whole bunch of different layers of things designed to make it difficult to copy.  These things did nothing to prevent the disc from being copied.  They did, however, make it hard for the customer who had actually paid for the disc, to listen to the disc.  For sure that's one person who's going to be looking a lot more closely at the next disc they're considering buying, and thinking twice about paying for one that has the "Copy Controlled" logo instead of the "CD Digital Audio" logo.  And on top of all that, EMI probably paid a whole lot of money to develop and/or license this technology, and billed it to the artists.  Everyone loses from this scheme except the pirates, and the techies who developed the copy protection scheme. Just think if all that money and effort could have been directed to something constructive instead.

Comments

Copyright © 2005, 2007 Matthew Skala
Updates to this entire site:
Updates to this category (copyright) only: