Bowie: EMI 1999 vs EMI SACD 2003

[krabapple]

'Up the Hill Backwards' CE stats from


1) Scary Monsters (CD)
"Digitally remastered by Peter Mew and Nigel Reeve @ Abbey Road Studios, London, 1999"
(inside back cover): "Digital remaster (c) 1999 "

Code: Select all

Min Sample Value:   -30106   -28704
Max Sample Value:   29850   29719
Peak Amplitude:   -.74 dB   -.85 dB
Possibly Clipped:   0   0
DC Offset:   -.001    -.001
Minimum RMS Power:   -96.66 dB   -97.28 dB
Maximum RMS Power:   -8.6 dB   -8.76 dB
Average RMS Power:   -16.37 dB   -17.67 dB
Total RMS Power:   -15.43 dB   -16.57 dB
Actual Bit Depth:   16 Bits   16 Bits

Using RMS Window of 50 ms

2) Scary Monsters (SACD - CD layer )
"Digitally remastered by Peter Mew @ Abbey Road Studios, London"
(inside back cover): "PCM and DSD stereo layers digital remasters (c) 2003 "

Code: Select all

   Left   Right
Min Sample Value:   -30106   -28704
Max Sample Value:   29850   29719
Peak Amplitude:   -.74 dB   -.85 dB
Possibly Clipped:   0   0
DC Offset:   -.001    -.001
Minimum RMS Power:   -96.66 dB   -97.28 dB
Maximum RMS Power:   -8.6 dB   -8.76 dB
Average RMS Power:   -16.37 dB   -17.67 dB
Total RMS Power:   -15.43 dB   -16.57 dB
Actual Bit Depth:   16 Bits   16 Bits

Using RMS Window of 50 ms


[krabapple]

Frequency profile comparison for the same tracks:

no differences from 20 Hz to 20 kHz.

[Rspaight]

I have the "Let's Dance" SACD but not the CD, but I suspect that's probably the case as well.

I believe the Ziggy SACD CD layer is different from the 30th Anniversary CD -- they corrected the reversed-channels error and some other glitches. Whether it's in fact a new *transfer* I don't know. Again, I have the SACD but not the CD (either the 30th Anniversary or the previous Virgin issue).

Ryan

[Xenu]

My (admittedly casual) ABing seemed to suggest that the SACD's CD layer was in fact different from both the old Virgin and new Virgin, but...

[krabapple]

The kicker for the Scary Monsters track is that they *don't* null out. I' m guessing I'm just not matching the samples up right. Can someone tell me how to do this?

[lukpac]

Very carefully.

[Xenu]

Find a really conspicuous peak and attempt to match 'em. That's really the only help I can give. Do they sync, at least?

[krabapple]

They appear to be exactly the same length, but when I invert+mixpaste on onto the other, you get a hollow-sounding version of the song, rather than a null.

How would I synch them?

[lukpac]

Zoom in really close (sample level) and line them up. It isn't that hard, really.

[krabapple]

I haven't bothered with nulls of plain CD vs CD layer of the SACD, but I have bothered to digitize the DSD layer and compare it to the CD layer. Result: the mastering is significantly different between DSD and CD layers -- the CD layer has 3-4 dB less dynamic range (but at least there's no digital clipping).

This links to a web page with the statistics and screen caps for 'Up the Hill Backwards' (DSD was digitized at 32/88.2, normalized, converted to 16/44.1; CD layer was just ripped and normalized):

http://www.m-ideas.com/sullivan/auditio ... kwards.htm

[Xenu]

Interesting. I wonder why they didn't just DSBM it down?

[krabapple]

Interesting. I wonder why they didn't just DSBM it down?

er..wha?

If you're asking why they made one louder than the other, rather than make a Redbook copy of the DSD layer, it's the same reason it was done for Dark Side of teh Moon and other hybrid CDs, I suspect.

[Xenu]

Yeah, i.e. "sabotage." It feels like a lot of effort for something so petty.

[lukpac]

Interesting. I wonder why they didn't just DSBM it down?

SBMD.

krab, could you digitize the CD layer just as you did the SACD layer? At least that way any possible changes due to the process would be the same across the board.

As I've mentioned elsewhere, I've seen strange behavior in situations like this. The original digital stream had a few dozen peaks at 0 dB, but when played back and recorded (making sure the highest peak was recorded at -0.1 dB or something), all of those peaks were not at the same level. The highest ones may have been at -0.1 dB, but others might have been at -1 dB. Long story short, the overall level was lower even though the highest peaks were the same.

[krabapple]

Interesting. I wonder why they didn't just DSBM it down?

SBMD.

krab, could you digitize the CD layer just as you did the SACD layer? At least that way any possible changes due to the process would be the same across the board.

I did that too-- captured the analog output at 88 kHz etc -- , I just didn't post it, because when normalized, it's virtually the same as the CD rip. I'll stick it up there later tonight.

Is 'SBMD' = super bit mapping?

As I've mentioned elsewhere, I've seen strange behavior in situations like this. The original digital stream had a few dozen peaks at 0 dB, but when played back and recorded (making sure the highest peak was recorded at -0.1 dB or something), all of those peaks were not at the same level. The highest ones may have been at -0.1 dB, but others might have been at -1 dB. Long story short, the overall level was lower even though the highest peaks were the same.

I often see slight 'shifting' of the pattern of peaks when I switch views from CD rip to CD analog waveforms. ..but IIRC the average levels weren't significantly different. I'll check when I post the analog capture.

One thing that's very important (though I doubt it accounts for what you see) is that in my player the 'speaker level' settings *have* to be at 0 dB trim/boost -- leaving them at 'fixed' results in a 6 dB boost, and in actual *clipping* for some sources.

For this round of transfers, I've made double sure that all the player's settings were set properly.

I also transferred Dark Side of the Moon...the whole album, SACD and CD layers. I may post about that soon.