Yes - CttE CD (Diament vs. Diament)

I recently scored a really old copy of Yes' _Close To The Edge_ that was pressed by Matsushita Electric Ind Co Ltd.

An EAC test of Siberian Khatru showed 1620 repeated samples with this and the US WEA pressed Atlantic, but the Japan pressing was slightly longer.

I think these discs sound the same, perhaps the Japanese is ever so slightly crisper, but I was curious to get opinions. So what do you think?

Both of these discs were pressed from the mastering by Barry Diament:

Sample 1:

Siberian Khatru (US Pressing) Atlantic A2-19133

http://www.zupload.com/download.php?file=getfile&filepath=31821

Sample 2:

Siberian Khatru (Japan pressing) Atlantic SD 19133-2

http://www.zupload.com/download.php?file=getfile&filepath=31828

 

I like the Japan pressing better. It sounds a little more laid back, clearer and smoother to me.

 

Definitely better drum definition

 

Hey! That Japanese made for USA Matsushita CTTE cd is the very first cd I ever bought. In early 1989, even before I owned a cd player, because it was and remains my favorite album. Glad to know I made a lucky purchase that day.

 

I'm not really set up to do critical listening on the computer, so I can't judge these two, but I've got an interesting Japanese pressing of this album -- 20P2-2053, ostensibly the Diament mastering, but with, apparently, some uncredited (at least in English) additional smiley-face EQ added. Not too heavy-handed, it just sounds a little too sludgy and in-your-face compared to the regular American pressing.

The best sounding CD appearance of material from this album I've heard is "And You And I" on the original Atlantic pressing of Classic Yes (which I learned about from one of Night-Gort Dave's lists). It must've used a better source tape -- listen to the cymbal crashes, there's just more of the sound on there, not a compression or EQ thing.

 

Barry Diament mastered that one also!

 

Folks, when you load them into a wav editor, synch the starts, trim them to the same length, then invert/mix paste, they flatline in both channels (digital silence). They're the same mastering with slightly different index points. Nothing to see here.

 

The A2 prefix I think also indicates a record club pressing if I'm not mistaken.

 

Nothing to *see* is the whole point, which I think you've missed. EAC showed them to be the same, as I stated. But did you *listen* to the clips?

The Japanaese pressing seems a *tad* (and I do mean just a *tad*) crisper. It could be my mind playing tricks, but I think the twang in Squier's bass sounds a bit crisper on the Japanese pressing. If it does, this simply raises the question about "jitter" and other factors that might cause of very tiny sound difference even in clips that show to be digitally identical.

They both sound good to me, it's just interesting food for thought.

 

Your mind playing tricks is the more likely cause than jitter, and I frankly can't think of any other causes for two files that are digitally identical, to sound different. But you can investigate that - cut and trim the files as I've done so that they start and stop at the same place, and run a blind comparison series using the software tools described below:

http://www.hydrogenaudio.org/forums/index.php?showtopic=16295

 

You hear no difference?

 

?

Now, how does that question follow from what I wrote? 'Hearing' a difference between identical sounds is nothing unusual, if you're consciously or subconsciously expecting a difference. You really want to rule out that common expectation effect, before pinning the cause on jitter.


The same thing would apply to me, if I said I heard a difference. So what value would that have? And if I said I didn't, would that convince you of anything, moreso than the actual bit-for-bit comparison data?

 

Okay, I'm better now. Carry on.

 

Right, so... do you hear a difference?

Whether you do or not will only be interesting to add to the discussion. That's the "value it would have."

I did a test using WinABX just a bit ago. Let me see if I still have the program open and can find a log...

Yep! Here it is:

And I used the ABXY test, btw.

Here are my results from a dozen listens:

A file: C:\Documents and Settings\1\Desktop\Siberian Khatru (US).wav
B file: C:\Documents and Settings\1\Desktop\Siberian Khatru (Japan).wav

21:24:49 1/1 p=50.0%
21:25:52 1/2 p=75.0%
21:26:19 2/3 p=50.0%
21:26:37 3/4 p=31.2%
21:27:22 3/5 p=50.0%
21:27:46 4/6 p=34.4%
21:28:13 4/7 p=50.0%
21:29:22 5/8 p=36.3%
21:29:35 6/9 p=25.4%
21:29:48 7/10 p=17.2%
21:30:50 8/11 p=11.3%
21:31:03 9/12 p=7.3%

They both sound about the same, as I had originally stated. They both sound a tad dull, but nice. And certainly better than the Rhino or the HDCD, IMO. It seems the Japanese pressing might be a tad crisper but it's too close to really call with any certainty.

 

Jeff, do you know if the YesYears CTTE (Gastwirt) mastering is the same as his CD of the album?

Thanks,
Chuck

 

Pioneer has this right: there's nothing to compare.

A general word on jitter /ABXing if I may: these files are the same... and as jitter can't exist when data is buffered (stored on a hard drive, etc), then by definition, wav files don't contain jitter. Unless your soundcard is acting up, there's no way, repeat, no way, a computer can play these files differently.

 

No. It's different.

 

Well, according to the test I ran with WinABX when I listened to the clips a dozen times, there's a 7% chance I was guessing. So, that would mean that there is about a 93% chance that I was able to detect a difference, even though EAC doesn't.

But I'm not really interested in disputing the idea that files that are digitally identical simply are digitally identical. These clips sound so close that even for an audiophile this is getting ridiculous to even care. But my WinABX test did prove interesting. I'll presume I got lucky and these clips sound the same, but I'm not 100% convinced.

 

Jeff,

this has nothing to do with EAC. Any wave editor with an inverion feature is good enough to check if the data of two audio files is identical.

If the two files are identical except slightly different starting and ending points, then they should sound absolutely identical on your system. If not, then your system acts unpredictable, or you were able to identify them because of the slightly different starting and ending points.

 

Andreas,

It has to do with EAC, because *that's* the program I used to test these. But any wave editor capable of comparing waves would be the same, of course. I didn't mention EAC for any other reason than to explain the source I personally used.

I wasn't able to identify them based on starting or ending points. I simply identified based on what I intitially described in the post that began this thread. In fact, at times I would get just a few seconds into a clip and click my answer as to whether it was the Japanese or US pressing based purely on sound. But I did some further testing later and didn't fare as well.

My system is very predictable.

 

Well, more trials could gain you *more* certainty.

 

There's more NR on the Yesyears box. Gastwirt changed his attitude towards NR between the box and the CDs...become more of a 'less is more' sort. IIRC this also affected his two attempts at the Hendrix catalog; the second has less NR than the first.

 

Thank you. I am diggin' that Run with the Fox song so much.

 

I will assume you decided in advance to do 12 trials, or else did not know your cumulative results during the test (because the test isn't valid if you simply watched your score during the test and stopped the test when it reached an 'acceptable' level to you).

I will also assume you edited the files to start at exactly the same point, so that you couldn't tell them apart simply by the slight delay of one vs the other.

p=.073 (your result) means there was a 7.3% probability of a Type I error* -- concluding there's a difference even though in reality, the null hypothesis (no difference) is true . Or in other words, even if there is really no difference, if you did 100 X 12 separate trials, by chance alone you'd score 9/12 seven times. For statistically significant difference, the standard acceptable threshold of Type I error is 5% (p=.05). Once you get *below* that, you can start to think about ruling out the null hypothesis (and get ready for publishing).

But to really nail it down, you want to increase the *power* of the test as well. This is function not only of the p, but also of the sample size (and other factors). Getting 18/24 (p=.011) will be more convincing than 9/12. Getting 19/24 (p = .003) will be hard to argue with.

(*Type II error is the probability that you will fail to reject the 'no difference' (null) hypothesis even though it is false (there is really a difference). Harder to calculate, since it depends on how big the actual difference 'really' is. *Technically*, if you fail to reject the null hypothesis
in any case, you shouldn't draw a yes/no conclusion about the reality of the difference... because as true believers will so often remind you, you can't 'prove' a 'no difference' the same way that you can 'prove' a difference.)

 

I have to be an annoying uber-Yes-snob here and say that the instrumental version on the B-side of the single ('Return of the Fox' ) is *much* better.