It's not how your pits are made

Eye diagram of a BPSK modulated signal

Although there has yet to be an announcement on Eminence's site, it seems that their next project [via] is going to be the orchestration of the Tenmon-Shinkai connection, a selection of pieces from 5 cm/s, Voices of a Distant Star, The Place Promised in Our Early Days, and She and Her Cat. This seems to be a pretty tenuous mix, and I struggle to imagine what half the orchestra would be doing if Eminence tries to stay aligned with Tenmon's sparse — and at times ultra-sparse — composition.

I think a chamber interpretation of Tenmon-Shinkai would have more success, but I have a bias. Generally speaking, I prefer Eminence's chamber projects to the orchestra stuff. Or maybe that's because Yuki Kajiura's music lends itself to smaller numbers of musicians.

At any rate, I'm willing to be surprised for 2800 yen which, for a CD album pressed in Japan, is actually a bit of a bargain. I coughed up more for THYME's first 9uality.

However, if Promise had a cheaper option in the form of a "regular" CD, I'd take that in a hearbeat over the HiQualityCD (HQCD). Why the press release dares mention this piece of junk is beyond me. Maybe because they know that people will think that anything with the word "quality" must necessarily imply a superior product.

It's a topic that should have been beaten to death even before HQCD's conception, but in reality will never be. I don't think that OST collectors will have encountered this attempt on their bank accounts, but I think it worth mentioning anyway.

You see, HQCD is suspiciously like the Super High Material CD (SHM-CD, pronounced sham) in construction, but probably modified enough to avoid having to pay royalties to JVC and Universal. Functionally, both are identical to the Blu-spec CD (BS-CD, pronounced *censored*).

As a matter of fact, they all play on old, boring, Red Book players. Could it be that these things are nothing more than pricier versions of boring, old, Red Book CD's?

Yes? Yes. Nod your head, please.

If you care to know why, my attempt at an explanation follows. Otherwise, remember that you should only pay more for shiny metal circles if a) there is no choice, or b) it is a remastered version.

Physical Layer

The two most important system parameters to a listener are unchanged. You still get 16 bits per sample, with a sampling frequency of 44.1 kHz. The only way a digital master could sound better on another disc format is if there was a higher sample resolution, a higher equivalent-sample resolution (for sigma-delta modulator formats), optionally coupled with a higher sampling frequency. HQCD and friends have neither, because they do not deviate from Red Book in this regard.

A better master may appear on one of these things but whoever decides to release two masters at the same time should be fired on the spot, for being an asshat. On the other hand, a better master may appear at a later date, so if you want to be less cynical, you could make the case that HQCD, SHM-CD, and BS-CD are to Red Book what Red Book was to vinyl.

Then again, a better master may one day appear on a boring old Red Book CD.

The only difference any of the three can claim is in the fabrication of the disc. Looking at the English pitch for HQCD, they claim that the transparent material used by HQCD (and SHM-CD) results in, essentially, a more consistent pressing of pits, such that they are all of equal length, or more equal than what is possible with regular polycarbonate.

Paired with a more reflective backplane of silver alloy, this is supposed to give better sound quality. How? By reducing data clock jitter, as evidenced by their bar graph. Jitter is down, great, but jitter affects sample phase error, which translates into read bit error rate. The act of reading is not the act of composing, performing, or mastering.

It's like claiming that because a ball-point pen doesn't leak ink all over the place, that you will write a better story. Uh, no. Your text could be readable from a meter away and still suck. It could be typed and presented on a screen in something that is not Comic Sans, and still suck.

Moving on to the second figure, we have a voltage graph of … something. Presumably the voltage across the photo-diode of a CD player for multiple pit to non-pit (land) transitions. Great, a 410 mV peak-to-peak swing! Better signal to noise ratio! The eye pattern opens up meaning less opportunity for indecision! Less indecision means less sampling error!

Except, that is not an eye pattern.

The image at the top of this entry is a real eye pattern appropriate for two-level modulation, where there is only one indecision level, as opposed to two. The HQCD "eye pattern" is really just the adjacent plot zoomed in, but what do I know?

As for the final figure, recall that the master is not the disc. Anyone who tells you otherwise is just pretending to be an audio or signal processing engineer on the internet.

Okay, but isn't a lower read error rate objectively better? Of course. I didn't say it was bad, just that it was irrelevant, and definitely not worth paying more money for. These days, if you are seeing a ton of read errors, it's because:

  • Your CD player is dying on you
  • Your ride is really bumpy and your player lacks anti-skip.

How often has either occurred to you in the past year? Be honest.

If you happen to be plagued by either of the above, simply buy a cheap CD player with anti-skip protection. High error rate suddenly disappears. So what problem is being solved by HQCD again? Other than the problem of extracting more money from your wallet?

But suppose that the reader randomly gets a bit or two wrong purely through thermal or shot noise. Hey, it could happen, albeit with something like a one in trillion possibility.

Will the audio quality degrade?

No.

But the reader got a bit wrong!

It could get ten bits wrong, and nothing would happen.

Really?

Really. No one suspects the error correcting code. Maybe that's because none of these marketing things ever bother to mention it, since it largely invalidates the product they're trying to push.

But suppose that some bit errors propagate through and fool Mr. Reed and Mr. Solomon, resulting in a, as in one, wrong 16-bit sample. Will you notice?

Assuming that your digital to analog converter is absolutely brick-wall perfect in frequency response, and there is no signal loss between the speaker and your brain, and there is no noise whatsoever, and you are sharp enough to detect a sample that lasts for 22.675 microseconds, then yes! Yes, you will notice the difference.

Which is to say, no you won't because, among other things, a truly noise-free environment can only be had at absolute zero, in which case the electronics would have long frozen over and likely shattered, the universe would have frozen over, and you would be dead.

Did I mention that your heartbeat is also a source of noise, so your heart would have to stop, in which case you would be dead?

If you must insist on listening to music using a CD player in the afterlife (as opposed to, say, having Chopin play for you in person), and are so unhappy with your one in trillion, 22.675 microsecond error, how about you listen to the track again? Probability says that you can expect an error about 6 hours later; a CD is 74 minutes.

Conclusion and Future Work

We have argued that you should not pay more for something that is typed because it will not make the writing any better, only easier to read. But since you already know how to read, marginally clearer text will not impact your reading experience.

And on the off chance that a word is missing a letter, or two, or three, you can quickly figure out what the word is, because you already know how to read. Your reading prowess is such that even a missing word will not prevent you from reconstructing a sentence. You are that pro.

P.S.

There is one real hypothetical advantage, a contradictory statement if there ever was one. The supposedly better input materials may make your disc less decomposable. I say may.

But don't you just rip your CD's and encode to FLAC anyway? If I were worried about disc rot, then I'd also back up to flash storage. If I were absolutely paranoid, then I would back up to two flash drives.