Peter Gabriel - Best Digital Masterings?

Discussion in 'Music Corner' started by tlake6659, Dec 22, 2008.

  1. tmtomh

    tmtomh Forum Resident

    Here are the DR ratings for the original CD, the 2002 SACD (PCM conversation of the stereo SACD layer), the 2012 24/48 digital download, and the 2o12 CD:

    '86 CD SACD '12 Digital '12 CD
    1. 11 10 9 6
    2. 11 11 10 7
    3. 11 12 11 10
    4. 12 10 10 7
    5. 12 12 11 9
    6. 12 11 9 6
    7. 12 11 10 8
    8. 14 13 12 10
    9. 12 10 10 7

    Album DR:
    12 11 10 8

    (I used the remaster track order, so for the '86 CD I moved In Your Eyes to the end so the DR numbers for each track would be apples to apples.)
     
  2. Paul H

    Paul H The fool on the hill

    Location:
    Nottingham, UK
    Excellent work.
     
  3. JoshM

    JoshM Forum Resident

    I suppose it depends on your definition of “compressed,” but I go into this at length in the article with graphs and DR and R128 numbers, and it doesn’t meet mine.
     
  4. JoshM

    JoshM Forum Resident

    I think relying solely on DR is misleading. In many cases, I think the R128 data corresponds more with how we hear things. But I think the best method is to look at DR, R128, and waveforms together (alongside subjective listening). That’s why I include a GIF comparing waveforms of the ‘86 CD vs the ‘12 download in my article and mention that, even though the DR numbers are a point or two lower for the ‘12, the R128 numbers are almost identical. Taken together, it’s clear that the ‘12 download is just as dynamic as the original CD. (It’s also a much, much better transfer.)
     
    tmtomh likes this.
  5. JoshM

    JoshM Forum Resident

    I emailed with Matt Osborne at Real World. He says there must have been a mix up with the files and that he’d make sure it gets fixed. He also said that anyone who got the wrong files should email Real World back and they’d make sure you get the 24/48 files.
     
  6. Paul H

    Paul H The fool on the hill

    Location:
    Nottingham, UK
    Forgive me, I haven't read your article but your comment implied that the 1986 mastering and the 2012 download mastering had the same dynamic range, which they don't. It just seemed to be somewhat misleading for anyone looking to this thread for information about the least compressed versions available. Whether the compression is significant enough to be of issue is, of course, dependent on the listener's ears and tastes. So all we have left is to offer factual information. I still argue that your post was factually incorrect.
     
    George P and The_Windmill like this.
  7. tmtomh

    tmtomh Forum Resident

    In my view it becomes semantics beyond a certain point. R128 (based on my understanding, anyway) is a good measure of loudness. By that measure, the original CD and 2012 digital-file remaster are nearly identical.

    At the same time, the DR meter is very good at revealing peak limiting, given that it measures the different between RMS level and the 2nd-highest peak for each track.

    Putting those two together, there is a good chance that @Paul H is correct, in the sense that the 2012 might very well have some very light peak limiting, which did not significantly impact the overall loudness or dynamics of the music - that would explain both the slightly lower DR numbers and the nearly identical R128 data when compared with the original.

    It's also worth noting that EQ differences can, by themselves, create 1-2dB variance in the DR meter's numbers. In particular, a little more bass energy in an otherwise unchanged mastering will almost always reduce DR by at least 1dB, often 2dB. Similarly, a treble boost can, all other things held constant, increased DR readings by 1-2dB. And differences in stereo separation also can impact DR meter readings.

    I'm not saying the 2012 digital mastering is simply EQ'd differently, with no peak limiting - there might very well be peak limiting, although it would have to be very light. Because if you compare the original CD with the 2012 digital remaster, the tracks' DR readings vary by between 0 and 3dB. Usually if peak limiting is the main culprit for reduced DR, the remaster will have DR readings that are more consistent in their degree of difference from the original. So I would speculate that the source transfers for both versions have different DR levels even before mastering (because of different stereo separation and/or different levels of bass or treble capture in the transfer - or possibly different generation tape sources for all we know). And I would further speculate that at least some of the DR difference is due to different mastering EQ - the remaster has more bass, which as noted above will invariably lower DR slightly. And it is also possible that the remastering job (or I guess the original mastering, or both) varied a lot track to track, meaning that some tracks might have peak limiting on them while others do not.

    We'll never know exactly what happened, but as far as the listening experience I agree with @JoshM - the 2012 digital download is best of breed for this album. (Although I hesitate to add that I do not interpret Paul's comments as necessarily saying the 2012 sounds compressed or bad - he's just rightly pointing out that there is evidence that suggests some peak limiting.)
     
  8. Mr Wensleydale

    Mr Wensleydale Forum Resident

    Location:
    Derbyshire, UK
    I’ve just had a similar mail from Matt, so fingers crossed!
     
    tmtomh and JoshM like this.
  9. JoshM

    JoshM Forum Resident

    With all due respect, my post and article are not “factually incorrect.” You’re taking one data point — the DR score, with all of its limitations — as “fact” and discounting other, equally (perhaps more) valid, measures completely. I care deeply about dynamics, but I try to present a multi-faceted analysis of dynamics in my pieces, not just DR scores. I also think dynamics aren’t the only important factor. Sometimes transfer quality, EQ, etc. are more important, and (as @tmtomh says below), often these are related to DR scores in ways that don’t indicate limited dynamics.

    I agree with your take. All of the things you mentioned affect DR vs R128 differently, as does amount of silence before songs, etc. Taking DR scores as our only data point and treating them as gospel is a mistake, IMO.
     
    The_Windmill and tmtomh like this.
  10. Paul H

    Paul H The fool on the hill

    Location:
    Nottingham, UK
    I was actually taking my own visual assessment of several waveforms into account as well. It seems to me that mild limiting is clearly visible in several tracks. The first minute or so of Sledgehammer is, for example, noticeably louder on the 2012 than the 1986, yet the end of the track isn't. Regardless of cause, it seems to me that the difference in dynamic range is clear to see.

    I agree with most everything else said in this discussion but, for my money, the claim that dynamically, the two are the same is still false. We'll agree to disagree. I'm happy for this thread to record the questioning of the veracity of your claim even if it isn't fully refuted.
     
    George P likes this.
  11. JoshM

    JoshM Forum Resident

    "False"? "Question the veracity"? "Refuted"? This is getting absurd.

    By your own admission, you didn't read my column, which not only discusses the DR scores and the R128 values for the '86 CD vs. the 25th download but also provides an animated GIF of the waveforms of numerous tracks from the '86 CD and 25th download back-t0-back so that readers can compare them. Then you take my comment in this thread about the dynamics of the 25th download vs. the 25th CD out of context and subject them to your own narrow, quixotic definition of "compressed."

    As I said in my column, most songs on the '86 CD are a point or two higher in DR score than those songs on the 25th download. But the R128 values are very similar, and in some cases songs on the 25th download have a higher R128 value than the same songs on the '86 CD! Moreover, the waveforms are very similar for the two, especially compared to a genuinely compressed mastering like the 25th CD.

    You mentioned "Sledgehammer." Let's look at "Sledgehammer" from the '86 CD (top) and 25th download (bottom):

    [​IMG]

    Now, if you think that those are significantly different, you're welcome to think that. I disagree. It's clear that there are a handful of stray peaks that are higher in the '86 than the 25th, but overall the waveforms are very similar (no "brickwall"-ing going on). Maybe the transfer on the 25th was just done differently, as @tmtomh suggested above. But let's say, for argument's sake, some light peak limiting was done on the 25th download. Might that have been done for a reason? What happens if we ask Audacity to highlight clipping for us?

    [​IMG]

    Ah! Now it seems like not having those stray peaks might've provide us with the upside of less clipping! (The 25th CD master, FWIW, is virtually a block of red it clips so much.)

    You're welcome to define "compressed" however you want. But when I write my columns, I try to consider multiple data points for judging dynamics, and I also use my ears. Beyond DR, R128, and waveforms, the key questions to me are: "Is this what most people would consider to be a compressed mastering? If the two masterings were played at the same volume, would a careful listener say that one was notably more compressed than the other?" For both of those questions, the answer on the 25th download is "no," at least as far as I'm concerned.

    If your definition of "compressed" is "any version with DR scores below those of the version with the highest DR scores," you're welcome to go with that. If that's what you care about, you can safely ignore my columns and much of the discussion on these forums! However, there's absolutely no justification for pretending that something like a DR score provides a simple, irrefutable black-and-white judgment and then, on the basis of that, to begin declaring things "true" or "false" and impugning people's integrity and the "veracity" of their claims (which, once again, you didn't even read).

    As discussed above, DR scores are affected by many things that aren't really indicative of compression, such as EQ decisions. Indeed, the '86 CDs, which all share the exact same mastering, differ slightly in DR (and R128) values simply because they have different peak levels. Moreover, if I take a song from either mastering, open it in Audacity, and do nothing but add a bunch of silence to the beginning, I can increase its DR score!

    DRs aren't the be-all and end-all of dynamics and compression. Nor is all compression bad. Compression is used at virtually every stage of recording, and when done well there's nothing wrong with it. Nor is limiting always bad. Nor are dynamics and compression the only thing worth considering when evaluating different masterings! As I said in my column, this was one of the easiest decisions of any of my columns, since (in my humble opinion and after many weeks of close listening and research, as discussed in my piece) the transfer quality and EQ of the 25th download is head-and-shoulders above previous versions'.

    If you like your original CD because it's the version with the highest DR score, by all means, enjoy it. (My column doesn't have the force of law behind it!) Indeed, going into this column, I expected the original CD to come out on top, and I still think it's a solid mastering. But, to my ears, the 25th download clearly tops it.
     
  12. Solace

    Solace Forum Resident

    Location:
    Brussels, Belgium
    A very informed and well-articulated argument. FWIW, just using my ears, I prefer the 25th download to the 1986 CD (and every other edition including the Black Triangle and - narrowly - the 2016 vinyl download) as it’s both clearer and has more oomph. The latter I usually associate with slightly lower DR readings, as proves to be the case here.
     
    tmtomh, JoshM and The_Windmill like this.
  13. Paul H

    Paul H The fool on the hill

    Location:
    Nottingham, UK
    Woah, woah. Hang on a minute. You're the one getting absurd here. You're using the word "compressed" to mean something along the lines of "suffering from compression" - at least, that's my impression - which is something I never claimed. You're also entering into a discussion of the merits of compression, all of which I agree with but none of which are relevant to the point at hand: which is simply whether or not some compression has been added to the 2012.

    All I refuted was your claim that the 2012 had not been compressed. Whether it's noticeable or not is irrelevant. Whether it's benefitial or not is up to the listener. I still consider that it's clear from the waveforms you've posted above that Sledgehammer, as an example, has slightly more compression in its 2012 incarnation than it does in the original mastering. That's got nothing to do with my "definition", it's just a fact, DR scores or R128 be damned. Just look at the waveform. If you'd said in your original post that there had been extremely modest compression that is likely unnoticeable, I'd have "liked" your post and made no comment.

    I wish I knew how to post images on this forum but, please, dig up Big Time from 1986 and 2012, put them in a new thread, create a poll and ask the good readers of this forum if they'd consider the 2012 to have been subjected to compression.
     
    Paul P. likes this.
  14. JoshM

    JoshM Forum Resident

    This gets back to my previous comment: I don’t think that when anyone on this forum refers to a version of an album as “compressed” they mean “has ever been touched by compression.” Instead, they mean “has had the dynamics noticeably squashed.” If we’re talking about the former, basically any mastering besides an unmastered transfer of a mix would qualify as “compressed.” If we extended backwards in the recording process, almost everything would qualify. (It’s also ironic that we’re discussing “Sledgehammer,” which, from the beginning, was mixed and mastered relatively “compressed,” since it was earmarked as the single.)

    Looked at on its own, I don’t think anyone would hear the ‘12 download and call it “compressed.” The yardstick you’re using, as said above, seems to be the DR values of the ‘86 CD. I’ve outlined why I think that’s misleading.

    So, yes, I’m not using “compressed” to mean “has been touched by compression.” I am using it more along the lines of “suffering from compression,” as you put it. I think that’s how virtually everyone uses the term when discussing masterings on these forums (and elsewhere), since the former definition is wholly unhelpful and often meaningless. (Would using tape count? A tube stage?)

    Once again, my article includes info on DR score, R128 values, and a GIF of waveforms (including “Big Time”). I’m not attempting to hide any info or mislead anyone, despite what you implied. I’m trying to do the opposite.

    My comment here that you objected to referred to the confusion (as noted in my article) that arose from the vague lineage of the ‘12 download. Judging from comments on old threads here and elsewhere, many people who bought the 25th box assumed the download had the same mastering as the CD, which is unambiguously “compressed.” A big chunk of my article was clearing up that confusion and trying to figure out who mastered the download, since the box set’s liner notes don’t make that clear.
     
    John Buchanan likes this.
  15. John Buchanan

    John Buchanan I'm just a headphone kind of fellow. Stax Sigma

    Josh, loved this very detailed comparison. Maybe the graphs' background could be white next time for better visibility.
     
    tmtomh and JoshM like this.
  16. Paul H

    Paul H The fool on the hill

    Location:
    Nottingham, UK
    Which is all well and good, Josh. But the issue of whether or not any mastering other than an unmastered transfer would qualify as compressed is irrelevant. That's a given. ALL music has some form of compression applied. But we're talking about how it appears in the public realm, on records and CDs and the like. The point I challenged was your simple assertion that the '12 had not been compressed in comparison to the '86 - that's all we have, we don't have access to master tapes. If you'd written a post that said something along the lines of "the CD mastering has been subject to considerable compression whereas, by comparison, the compression applied to the hi res download is so modest as to be virtually undetectable" or similar, I'd have had no qualms. But you made a flat out statement that there had been no compression in comparison which I took objection to.

    When I find I moment, I'll read your article. I owe you that at least. Perhaps we're talking semantics, but on a forum dedicated to helping discerning listeners, I think one needs to phrase one's wording very careful sometimes. Heaven knows, there's enough misinformation on this forum already...
     
    George P likes this.
  17. captainsolo

    captainsolo Forum Resident

    Location:
    Murfreesboro, TN
    I was looking for a thread about digital versions of So and PG on cd in general.
    So I take it the original cd is the same mastering worldwide and both it and the download version are good choices?

    Ultimately does either one tame the heavy sibilance of the otherwise nice US artisan LP and most other vinyl editions?

    And is there any album that is generally recommended on a digital form over vinyl? I’m interested in getting UK original vinyl and currently have two great US originals: Mercury of Melt and a Quiex Townhouse Security.
    But I have no idea about his material on cd.

    Thanks for any info!!
     
    c-eling likes this.
  18. JoshM

    JoshM Forum Resident

    We’re just going circling back to the beginning now. Your initial argument for my comment being “false” was based on DR scores, and I explained why that’s misleading. Then it switched to waveforms, etc. The frustrating part is that all of this was covered in my article.

    Of course this is semantics. If you hadn’t framed my comment as “false,” I wouldn’t have even bothered responding. Despite people’s reliance on DR scores as gospel, there’s not a single objective measure of compression. When people refer to masterings as “compressed,” especially on audiophile sites, they do tend to mean it as a criticism — that the recording’s dynamics have been noticeably reduced (“loudness wars,” etc.). My point was that, despite the misleading DR scores, the ‘12 download has dynamics that are very similar to those of the original CD. I stand by that as being accurate. If you disagree, that’s fine. But don’t go throwing around things like “false” and “veracity” and “refuted.”

    For further correspondence, please consult Lionel Hutz:

    [​IMG]
     
    LivingForever and tmtomh like this.
  19. c-eling

    c-eling They're made of light,We never would have guessed

    IV I enjoy the old German mastering, not sure if the old US is just a level shift or not. But the target sounds more 'natural' to my ears. III I haven't done any serious back and forths with the old US DADC and UK PG3 Nimbus. They do sound different however. I, I enjoy my old US SRC, haven't gone further. II I only have the old US LP.
    Weird Solo, I prefer Gabriel on old compact disc than vinyl for some reason.
     
    captainsolo likes this.
  20. The_Windmill

    The_Windmill Forum Resident

    Location:
    Italy
    No. There'e one version that's more widespread (the good one) and another that was pressed in Japan and USA (crappy).
    The "good" one comes in different level-shifted versions.

    See Peter Gabriel - Best Digital Masterings? and following comments
     
    Last edited: Jun 10, 2019
  21. Paul H

    Paul H The fool on the hill

    Location:
    Nottingham, UK
    Thing is, Josh, it wasn't. I made no comment about why I thought your comment was misleading. I just said I thought it was wrong. Someone else posted DR scores and you latched onto it. You've never actually bothered to ask me why I thought your observation was erroneous. Read the discussion again if you don't believe me.

    If, by your own claim, there is no single objective measure of dynamic range, then we have to use the tools at our disposal which include listening, looking at waveforms and comparing DR values. Let's say, for the sake of argument, we achieve the following result:

    Listening: no appreciable difference
    R128: no appreciable difference
    DR: noticeable difference
    Visual inspection of waveform: noticeable difference

    I'd argue that "a jury" would find no compelling case either way: it can't be proved beyond doubt that there's no difference, nor that there is a difference. Put another way: there's sufficient doubt to give cause for questioning this issue.

    Either way, it offers enough doubt for me to feel that your assessment shouldn't be taken as an absolute, as a fact. Which is how you framed it.
     
    Last edited: Jun 10, 2019
  22. JoshM

    JoshM Forum Resident

    I am primarily concerned with how things sound, even if I don’t rely on listening as the sole data point (something, once again, made clear in my piece), since that’s ultimately what matters in audio, IMO. But I never — not in my article, not in my comments here — called my views “absolute” or “facts.”

    I only responded to you because you claimed that what my comment “implied” was “factually incorrect” based on “dynamic range”:

    By your own admission, did this all without reading my article, which contained all of the data discussed — DR, R128, waveforms — and made my meaning clear (no discerning what I “implied” necessary).

    Instead of just saying that you misunderstood me based on a mistaken assumption of what I meant by “compressessed,” you’ve resorted to increasingly pedantic and lawyerly responses (hence my Lionel Hutz joke), which you only underscored with your last comment.

    Based on my research, using multiple data points, I don’t think the ‘12 download is “compressed” in the common meaning of the term on this site and elsewhere. Indeed, I don’t think there’s any practical difference between it an the ‘86 CD in terms of dynamics. I explain this in my article, without hiding anything (including DR scores), despite your claiming I was spreading “misinformation.” But I certainly never asserted anything as an absolute fact.

    I write my pieces out of a genuine passion for the albums and a desire to try to help consumers find the best available masterings. I do the best I can at that, providing as much info as possible and making my reasons transparent, with no ulterior motives.

    You’re welcome to disagree, and you’ve register your disagreement. I honestly don’t care if you disagree, but don’t claim that I’m lying or spreading misinformation. That’s what I objected to.

    “Case closed.”
     
    LivingForever likes this.
  23. JoshM

    JoshM Forum Resident

    To my knowledge, there's only one mastering for all of the '86 So CDs. From the list above, I compared CDs with track one/two peaks of 100/100, 98/99, and 75/75 in Har-Bal. All were the same mastering, just level shifted. (If someone has the Canadian CD, with track one/two levels of 100/97, send me a message, and I'll check it, too, just to be sure.)
     
    tmtomh likes this.
  24. HiFi Guy 008

    HiFi Guy 008 Forum Resident

    Location:
    New England
    Yet, some of those same masterings (at least they have the same peak levels) sound very different from each other to my ears.
     
    The_Windmill likes this.
  25. The_Windmill

    The_Windmill Forum Resident

    Location:
    Italy
    Then I suppose your knowledge doesn't include the last one in the list, US Geffen 9 24088-2, which is different and worse.
    BTW, there are more informations in a post slightly above the one I linked.
    Both software and listening comparisons confirm they are different. They sync but have different EQ (and maybe something else going on I can't tell...).

    All the other masterings are derived from the same one (The Canadian MCA too, we can assume), same clipping but different levels, BUT the 75% might perform well on some systems (like @HiFi Guy 008 's :wave: ), my guess is because it avoids intersample distortion and could be taking more advantage of the amp's performance.
     
    Last edited: Jun 11, 2019

Share This Page

molar-endocrine