Pink Floyd - Dark Side of the Moon - Analysis

Discussion in 'Music Corner' started by foobar2000, Oct 6, 2011.

  1. foobar2000

    foobar2000 New Member

    Location:
    US
    Pink Floyd - Dark Side of the Moon - Analysis

    OK, I am going to post some technical analysis of the various masterings of Dark Side of the Moon, including the new one just released.

    99% of you will find this boring as all heck, so just ignore me. Just to head off any comments; Yes, I listen to the music too. No, I won't be posting any comments about which mastering sounds more buttery. (Feel free to do so yourself, however.) Yes, I find this sort of analysis fascinating in its own right. Yes, I am a hopeless geek.

    So, with that out of the way. . . .
     
  2. foobar2000

    foobar2000 New Member

    Location:
    US
    There are seven distinct digital masterings of Dark Side of the Moon. 1) The 1983 original CD with pre-emphasis, probably mastered by Sony in Japan. 2) The 1984 CD without pre-emphasis (The dreaded "TO" in the matrix version.) probably mastered by EMI in Germany or perhaps Toshiba in Japan. 3) The 1985 made-in-USA Capitol, perhaps a tweaked version of #1? 4) The 1988 Mobile Fidelity Sound Lab gold CD. 5) The 1993 to start off the band-approved re-masterings, done by Doug Sax, also released with slight variations in the Shine On boxset. 6) The 2003 dual-layer SACD, done by James Guthrie, the redbook version of which replaced the standard European CDs. And 7) The brand shiny new one, released last week, in both hi-rez and lo-rez.

    Peak values are as follows:

    1983 - Japan Toshiba, Eu Harvest, US Capitol
    Sony Mastering, with pre-emphasis (non-TO)
    39.9 - 43.3 - 70.1 - 55.2 - 72.3 - 44.5 - 53.9 - 54.6 - 61.2

    1984 - Japan Toshiba, Eu Harvest, US Capitol
    EMI and/or Toshiba Mastering (TO)
    87.8 - 94.7 - 100 - 97.2 - 100 - 99.1 - 100 - 96.1 - 100

    1985 - US Capitol
    Capitol Mastering?
    46.8 - 48.2 - 49.3 - 51.0 - 72.5 - 63.3 - 67.1 - 58.8 - 62.7

    1988 - Mobile Fidelity Sound Labs
    MOFI Mastering
    29.3 - 34.1 - 39.2 - 40.3 - 41.5 - 56.3 - 49.2 - 57.8 - 52.9 - 60.0

    1993 - Japan Toshiba, Eu EMI, US Capitol
    Mastered by Doug Sax (with James Guthrie) at The Mastering Lab, 1992
    67.9 - 69.1 - 76.9 - 75.0 - 96.8 - 69.6 - 89.3 - 75.7 - 82.5

    2003 - Japan Toshiba, Eu EMI, US Capitol (SACD)
    Oh by the Way boxset, & Eu EMI CD releases
    Mastered by James Guthrie at das boot, 2003
    34.4 - 92.4 - 92.0 - 93.0 - 92.7 - 100 - 84.3 - 100 - 100 - 100

    2011 - Discovery, Experience, & Immersion worldwide releases
    Mastered by James Guthrie and Joel Plante at das boot, 2011
    40.6 - 92.8 - 99.5 - 96.6 - 98.4 - 100 - 90.6 - 100 - 99.7 - 100
     
  3. foobar2000

    foobar2000 New Member

    Location:
    US
    I'm going to concentrate on two tracks, Money and Time.

    First, Money.

    I trimmed all the versions to start at the exact moment the bass guitar comes in, and end exactly five minutes later. This was to avoid any differences in fade-ins or index points, yet give me a full song-length clip to analyze.

    These are all "proper" rips, using Exact Audio Copy for the regular CDs. The 1983 CD was equalized to remove the pre-emphasis using SoundForge and the iZotope Mastering EQ plug-in which has a "Post-Emphasis Curve" pre-set. The DSD files were ripped off the SACD, without any conversion to PCM or analog, using a PS3. The DSD files were then converted to PCM using the Korg AudioGate. Korg recommends reducing the gain when converting due to the fact that DSD can produce a louder sound than PCM can. Indeed when I did a straight conversion I got lots of clipped peaks. When I reduced the gain by 2.5dB the loudest peak was -0.1 dB, and I saw no squared off tops there. So, what you see here may not be exactly what is encoded on the SACD, but I believe it's as close as we can get in PCM. The Blu-Ray was ripped using a combination of DVDFab and eac3to.

    The absolute phase is reversed on some of the releases. #1, 3, & 6 are the same - #2, 4, 5 & 7 are the same. I have no way of knowing which is correct.

    Here are the ReplayGain values and TT Dynamic Range Meter scores:

    Money - 1983 Sony
    RPG Track Gain: +0.62 dB
    RPG Track Peak: 0.564728
    TT DR Meter Score: DR11

    Money - 1984 EMI
    RPG Track Gain: -6.46 dB
    RPG Track Peak: 1.000000
    TT DR Meter Score: DR11

    Money - 1985 Capitol
    RPG Track Gain: -1.19 dB
    RPG Track Peak: 0.725067
    TT DR Meter Score: DR11

    Money - 1988 MFSL
    RPG Track Gain: -0.47 dB
    RPG Track Peak: 0.563416
    TT DR Meter Score: DR10

    Money - 1993 Sax
    RPG Track Gain: -5.18 dB
    RPG Track Peak: 0.968201
    TT DR Meter Score: DR10

    Money - 2003 Guthrie-RedBook
    RPG Track Gain: -7.97 dB
    RPG Track Peak: 1.000000
    TT DR Meter Score: DR8

    Money - 2003 Guthrie-SACD
    RPG Track Gain: -6.00 dB
    RPG Track Peak: 0.994568
    TT DR Meter Score: DR9

    Money - 2011 Guthrie-RedBook
    RPG Track Gain: -6.04 dB
    RPG Track Peak: 1.000000
    TT DR Meter Score: DR10

    Money - 2011 Guthrie-BluRay
    RPG Track Gain: -4.29 dB
    RPG Track Peak: 0.961517
    TT DR Meter Score: DR11

    Note that these are for the trimmed down five minute clips, not the whole song. Also that the 1983 Sony is in de-emphasized state, which reduces the volume.
     
  4. foobar2000

    foobar2000 New Member

    Location:
    US
    Here are what the waveforms look like:

    Money 1983 Sony
    [​IMG]

    Money 1984 EMI
    [​IMG]

    Money 1985 Capitol
    [​IMG]

    Money 1988 MFSL
    [​IMG]

    Money 1993 Sax
    [​IMG]

    Money 2003 Guthrie RedBook
    [​IMG]

    Money 2003 Guthrie SACD
    [​IMG]

    Money 2011 Guthrie Redbook
    [​IMG]

    Money 2011 Guthrie BluRay
    [​IMG]

    (You wanted waveforms, right?)

    Once again, these are my five minute clips, and shown in SoundForge.

    You can see from this (and the numbers above) that the two Guthrie masterings are not the same in their redbook and hi-rez versions. The 2003 redbook being significantly clipped, and the 2001 slightly so.

    From an overall volume we can divide them into three groups; 1) The two clipped Guthrie redbooks. 20 The two Guthrie hi-rez, Sax, and 1983 EMI. 3) The 1983 Sony, 1985 Capitol, and MFSL. Obviously going over 0dB reduces the dynamics, but so do the very quite ones by staying well away from 0dB. The middle group (in theory) are the ideal.
     
  5. foobar2000

    foobar2000 New Member

    Location:
    US
    Now for the fun part; Spectral Analysis.

    I import each of my five minute clips into Audition, highlight the full length, perform a spectral analysis, and export the data. I then import the data into Excel, and create a chart.

    Like so:
    [​IMG]

    You can't really see much there, right? To make the picture clearer I use one mastering as the "baseline" and do a subtraction of all the others.

    Like so:
    [​IMG]

    This essentially straightens out the red line (the 1983 Sony) to show us how the other masterings differ. This does not mean that the Sony was "flat" and the others "wrong" just that (for instance) the 1984 EMI has a lot more spectral energy around 7k hz then the Sony does. Nor does it tell us how it got there - it could be EQ, tape machine alignment, the source itself, etc.

    It does paint a very accurate picture of what the spectral balance is however, never mind how it got there. For example when I compare the raw 1983 Sony file to the de-emphasized one I get a perfect pre-emphasis curve. So, this will tell you something important about how these masterings sound. (Not everything, but something.)

    Here is the 1993 Sax as baseline:
    [​IMG]

    And zoomed in with just the Sax and Guthrie masterings:
    [​IMG]

    You can see that no matter which one I use as the baseline certain characteristics show up. That 7k peak in the 1984 EMI. The MFSL "smiley face". The enormous high-end boost in the Capitol. (But, see the next batch.) Comparing the two Guthries you can see a slight (1dB) high- and low-end boost on the newer one.

    When I compare the two Guthrie hi-rez to their redbook versions, meaning subtract the 2003 SACD from the 2003 redbook, I get a straight line. This means the spectral balance is exactly the same, EQ "moves" were exactly the same, and the only differences are in the dynamics. Same goes for 2011 blu-ray vs 2011 redbook. The ones shown in the chart are the hi-rez, but it would look exactly the same if I used the redbook versions, just the lines would be a couple dB higher.
     
  6. Muzyck

    Muzyck Compulsive Completist

    Whoa! Vernon eat your heart out ;)
     
  7. foobar2000

    foobar2000 New Member

    Location:
    US
    Bored yet?

    Here is Time. From the instant the alarms sound, to exactly five minutes later. Again this is to eliminate differences in index points.

    Time - 1983 Sony
    RPG Track Gain: +1.28 dB
    RPG Track Peak: 0.397339
    TT DR Meter Score: DR9

    Time - 1984 EMI
    RPG Track Gain: -5.99 dB
    RPG Track Peak: 1.000000
    TT DR Meter Score: DR10

    Time - 1985 Capitol
    RPG Track Gain: +0.75 dB
    RPG Track Peak: 0.493561
    TT DR Meter Score: DR10

    Time - 1988 MFSL
    RPG Track Gain: +1.00 dB
    RPG Track Peak: 0.403198
    TT DR Meter Score: DR9

    Time - 1993 Sax
    RPG Track Gain: -3.90 dB
    RPG Track Peak: 0.769928
    TT DR Meter Score: DR9

    Time - 2003 Guthrie-RedBook
    RPG Track Gain: -6.03 dB
    RPG Track Peak: 0.930267
    TT DR Meter Score: DR9

    Time - 2003 Guthrie-SACD
    RPG Track Gain: -4.06 dB
    RPG Track Peak: 0.803253
    TT DR Meter Score: DR10

    Time - 2011 Guthrie-RedBook
    RPG Track Gain: -5.60 dB
    RPG Track Peak: 0.966919
    TT DR Meter Score: DR10

    Time - 2011 Guthrie-BluRay
    RPG Track Gain: -3.85 dB
    RPG Track Peak: 0.791443
    TT DR Meter Score: DR10

    Note how the TT DR scores are either 9 or 10 here, not much difference.
     
  8. foobar2000

    foobar2000 New Member

    Location:
    US
    The Time waveforms:

    Time 1983 Sony
    [​IMG]

    Time 1984 EMI
    [​IMG]

    Time 1985 Capitol
    [​IMG]

    Time 1988 MFSL
    [​IMG]

    Time 1993 Sax
    [​IMG]

    Time 2003 Guthrie RedBook
    [​IMG]

    Time 2003 Guthrie SACD
    [​IMG]

    Time 2011 Guthrie Redbook
    [​IMG]

    Time 2011 Guthrie BluRay
    [​IMG]

    Although we don't get the same clipping on this track you can see the same basic groups as before. One thing to note is how much more dynamic the alarms are in one channel of the 1984 EMI. I wonder if that's the way the tape is, and everyone else has reduced the volume over the years, or if it's something the engineer did for this release?
     
  9. ssmith3046

    ssmith3046 Forum Resident

    But which one is the most buttery?
     
  10. foobar2000

    foobar2000 New Member

    Location:
    US
    And, finally, the Time spectral analysis:
    [​IMG]

    With the 1983 Sony as baseline:
    [​IMG]

    With the Sax as baseline:
    [​IMG]

    And just the Sax and Guthries:
    [​IMG]

    Looks pretty similar, suggesting that whatever caused the spectral balance to change is the same across the whole album. (No, I can't do every track, this takes to long.)

    The one that does change drastically is the 1985 Capitol. I've seen this before on those early US Capitol discs, specifically Wish You Were Here. Big swings in spectral balance. I wonder if we're actually looking at an error when they tried to release a disc without pre-emphasis? The curve in Money sure looks like the emphasis curve. Or maybe they just decided Americans would like a 10dB boost in the highs?

    Well, hope that was somewhat informative, to some one.

    There will be a test next week. . . .
     
  11. foobar2000

    foobar2000 New Member

    Location:
    US
    I should mention that this wouldn't have been possible without using Vernon's awesome website as a reference.
     
  12. foobar2000

    foobar2000 New Member

    Location:
    US
    The West German EMI, of course!

    (Japanese don't use butter.)
     
  13. First, let me thank you for doing all this work. I'm aware of how much work this took to create, especially including the DSD mastering from the SACD. It's exactly the type of information I was looking for about the new Pink Floyd remasters. These graphs cut through a lot of the endless back-and-forth discussion that tends to dominate the reviews.

    So how soon can we see the Wish You Were Here comparison?:D Hopefully after the the SACD is released...
     
  14. foobar2000

    foobar2000 New Member

    Location:
    US
    I did Wish You Were Here a while ago. There are not nearly so many versions, no MFSL, no 2003 Guthrie. Maybe time to re-visit it. As you say after the SACD, and the hi-rez Immersion, comes out.
     
  15. testikoff

    testikoff Well-Known Member

    I think you may have needed to normalize tracks (say, by applying respective track ReplayGain values) before any spectrum comparisons...
     
  16. Ramos Pinto

    Ramos Pinto New Member

    Location:
    Southeast US
    re: "Time"

    Judging from post #10, the 1983 Sony would be. :thumbsup:
     
    Dave likes this.
  17. foobar2000

    foobar2000 New Member

    Location:
    US
    That would not do anything at all besides move the lines up or down the graph, not change the curves in any way. Normalizing as no effect on the spectral analysis.
     
  18. foobar2000

    foobar2000 New Member

    Location:
    US
    How so?
     
  19. GreenDrazi

    GreenDrazi Well-Known Member

    Location:
    Atlanta, GA
    Great work! Thanks
     
  20. LarsO

    LarsO Well-Known Member

    For me it was very interesting to see the last spectrum curves with Sax as a reference since that's my only (on CD) reference. Some of my first thoughts listening to the new one was that there was more power in the bottom and high end and I'm happy to see that proven. Thanks for your effort!
     
  21. marcel

    marcel Alea Iacta Est

    Location:
    Italy
    Thanks for sharing, Foobar! :thumbsup:
     
  22. coffeecupman

    coffeecupman Forum Resident

    Location:
    Austin, Texas, USA
    Great post. I'm going to have to try that EMI sometime. Serious mids party.

    ccm
     
  23. testikoff

    testikoff Well-Known Member

    Your spectrum differences curves will be more meaningful this way, IMO. It's not just about the shape of the spectrum, is it?
     
  24. rockitman

    rockitman Active Member

    The vinyl versions of course. probably the Japanese EMI Pro-Use version followed closely by the Mobile Fidelity (Japan press version). ;)
     
  25. ctyankee

    ctyankee New Member

    Location:
    Wilton, CT USA
    Foobar,

    Lots of charts, yes, but what's the takeaway from the effort?
     

Share This Page