Oh for f*cks sake people. I don't mind pedantry - this entire topic is all about splitting hairs - but this is getting a little weird.
QUOTE (krabapple @ Jun 30 2010, 00:56)
I think most of us can agree:
-- vinyl masterings are OFTEN -- maybe TYPICALLY -- different from the corresponding CD release
-- but by no means NECESSARILY different
And that these are the most salient points.
I can agree with these points, buuuut... I really am asserting something much stronger than this.
I don't claim to have much of any hard evidence for my beliefs here, besides counterexamples.. Such is the life of of the amateur researcher. But I will
claim that there isn't any hard evidence arguing the contrary position. And the soft evidence I do have is pretty damn strong TYVM.
So here I will try to outline precisely
what I am trying to argue here, as a hypothesis, which is both quite falsifiable and not impossible to find hard evidence for assuming it's true.
For the class of vinyl releases with a "hypercompressed" CD of the same material released simultaneously, where there are no public claims one way or the other about the processing chain of the vinyl master (ie, at which point does the signal flow of the CD and vinyl masters diverge): let's say that some proportion of these vinyl releases' masters are sourced from a signal less hypercompresed than the CD master. Let's call this proportion Plh
(As a point of reference, the "myth" claim that analog scott originally objected to, I believe, represents the belief that Plh=1
- that, due to restrictions on modulation velocity/acceleration, cutting head temperature limits, tracking etc, no
hypercompressed material is allowed on vinyl. I think krab's post settled everybody on that point.)Hypothesis: Based on all available evidence, Plh << 0.5.
That is, for albums with simultaneous CD/vinyl releases, where the CD is hypercompressed and the vinyl is of unknown mastering provenance, one would expect very few vinyl masters to be less hypercompressed then the CD masters. Therefore, buying a vinyl release, in order to (primarily) avoid hypercompression that is known to exist on the CD, is not
likely to do so.
- There is no conclusive empirical evidence supporting the position that Plh >= 0.5. AFAIK, there isn't even evidence that Plh > 0.5 in any reasonably sampled subset of vinyl releases that fit the class of releases being discussed - like, what one would find in a given record store, or for a particular genre or even label.
- There is much anecdotal evidence supporting the position that 0 < Plh < 0.5. In fact, from anecdotal evidence, Plh ought to be predicted to be fairly close to 0. I base this claim on the following points.
- Based on my own (very limited) analysis comparing digital transcriptions of vinyl to CD versions that I have access to, every hypercompressed CD I looked at looked clipped to exactly (or almost exactly) the same magnitude on the transcription. IIRC, this is for perhaps a sample size of roughly 2-3 albums. For albums with mild or nonexistent hypercompression on CD (a much larger number), all of the equivalent vinyl transcriptions I've looked at had no additional dynamic range and no unambiguous improvement in limiting/clipping magnitude.
- The impression I get, from reading comments made by cutting engineers on forums, is that the overwhelming majority of the sources they receive for vinyl mastering are the same as the equivalent CD masters.
- A priori, there are reasons for believing the near-certainty that Plh < 0.5, and there are decent reasons to believe Plh << 0.5.
- My understanding of the vinyl mastering process is that, overall, it is typically more expensive or more time-consuming to source the vinyl master from a separate audio signal than the CD master, particularly if it is to be meaningfully separate (ie, does not contain hypercompression present on the CD release).
- If extra attention was paid to the vinyl mastering, in the sense that it was less hypercompressed, assuming this represented an added expense in production, it would be reasonable to expect that this would be prominently featured in the marketing for the vinyl release, to best capitalize on the investment. This was most certainly true of, eg, Steve Hoffman's mastering of Icky Thump[i], of the mastering of [i]Mudcrutch, etc. I have no idea what proportion of Plh for which this is actually true, though.
- My impression of the opinions of "a significant fraction" of people - perhaps "most" - who prefer vinyl over CD on sound quality grounds, are going to base their belief more on the intrinsics of the formats, rather than the quality of the masters being used. (It would not surprise me if most producers/musicians with such preferences believe the same thing.) From a return-on-investment point of view, this is an incentive for a label not to use a less hypercompressed master for vinyl, because such a feature simply would not sway the buying decisions of some/most customers and thus not provide a significant return on investment. It is worth noting that the belief that hypercompression is inimical to vinyl as a format plays into this incentive.
- Fundamentally, this is not a sound quality that consumers have terribly accurate accuracy in estimating. IIRC, Bob Katz estimates in Mastering Audio that transients can usually be hard limited to -6db down without audible insult. Moreover, even trained and well-respected mastering engineers will prefer vinyl even when it is hypercompressed to the same degree that the CD is. And most music listeners only know about the "loudness wars" only in the context that their CDs mastered in the 1980s sound much quieter than those in the 90s/2000s - sound quality itself tends to be a secondary discussion. That leads me to believe that, while hypercompression in itself is often audible and ABXable, believing differences in its magnitude are consistently identifiable is very unjustified.
This matters because IMHO product quality in a market is positively correlated with the accuracy in which consumers are able to estimate it. When there is little reason to believe that consumers will consistently tell the difference when a vinyl master is less hypercompressed than a CD master, there is little reason to believe that most labels and producers will go to such troubles.
By "hypercompressed" here, I mean that multiple points exist in a 44.1khz PCM signal, where there is constant (or nearly constant) slope for over 0.5ms (22 samples) of the waveform, and such constant slope is best explained by aggressive hard limiting or outright clipping at some point in the signal flow (from recording to mastering). A signal which is significantly "less hypercompressed" than another will have correlated peaks but much smaller durations of constant slope.
OK - does that make sense to everybody? Is this a reasonably cogent argument? How may this be full falsified (or proven)?
But out of humanitarian concern Axon should modify the wiki language to prevent a tragic twisted-underwear-induced injury to sensitive readers.
Point taken, and after this argument is over with I am certainly going to rewrite the page in places, including with this issue. But there are some pretty fundamental issues people are having with the page as a whole that need to be discussed first.