Normally I'm not a huge fan of crying on a forum for help but I'm at my wit's end on this one.







These are CD vs vinyl waveform plots. The CD plots are the ones with the razor-flat horizontal clipping lines. The vinyl plots are the ones with the sloped lines at the same locations

What causes this sloping?

I'm having a really hard time coming up with an agreeable explanation. Originally I thought it was rumble, but it is occurring at too high a frequency (coming and going in the course of under 10ms) to be that. A significant HF boost could cause this but a comparison of the spectral content of cd vs vinyl does not show anything like that happening. Some sort of allpass filter might explain this (that some frequencies just have a far different group delay between formats), but I asked at PSW if they've ever heard of using an allpass in mastering, and one vinyl engineer shot that idea down cold.

This is an extremely important question because some mastering engineers are using these waveform plots as supporting evidence to show that the vinyl masters are different from and superior to the CD masters. That is, their arguments are that: the peak content is higher in general on the LPs; and in the clipping regions, the "attack" part of the clipping is almost invariably higher in magnitude than the later "decay" portion, so that more transient content is being preserved on the LP.

I think those arguments are bollocks - the duration of the clipping is the same between the masters, so how can one be intrinsically less clipped than another? - but nobody seems to be buying that argument. Of course, if this position on vinyl superiority were to be accepted, virtually every vinyl record out there would be shown to be superior, because virtually all of them (in my experience) exhibit this kind of sloping. And IMHO that's a fundamentally wrong position to take on mastering that will needlessly waste listener money and lead to regressions in sound quality.

Am I missing something fundamental here? If not, what kind of evidence can I show that these vinyl masterings can be obtained by applying distortions to the CD masters?

I may be totally wrong but the shape of the vinyl master looks like a digital limiter is doing its work. Maybe you find similar shapes in other cds.
If i remeber right Amy Winehouse´s Back To Black has similar shapes in. I once looked at the waveforms out of interest how that loud recordings look like
Since the vinyl of Metallica is from a digital source they used a different mastering approach from the beginning and therefore a different type of limiting.

>What causes this sloping?

The application of a high-pass filter (anti-rumble?) to a clipped waveform.

IT IS NOT GETTING HIGHPASSED. At least, if it is, it's less than 2db. And I really doubt that 2db, at any corner frequency or Q, would cause that.

How do you know it isn't getting high-passed?

If it walks like a duck...

As a non expert at all (but having had LP's at home and even unintentionally broken some of my brother's ones when i was a child.. ), I wonder....

Could it simply be the needle?

I theorize: A needle will always move to the centre of the waveform, for any DC offset, since it has no knowledge of where that DC is (there's just one needle after all).

How do you know it isn't getting high-passed?

If it walks like a duck...

Spectrum comparison, LP response divided by CD response, 50ms window, 1/2 overlapped, left and right channels:

It's not being highpassed.

' date='Oct 9 2008, 16:11' post='592610'] As a non expert at all (but having had LP's at home and even unintentionally broken some of my brother's ones when i was a child..  ), I wonder....

Could it simply be the needle?

I theorize: A needle will always move to the centre of the waveform, for any DC offset, since it has no knowledge of where that DC is (there's just one needle after all).

This is happening on a pretty wide range of playback systems, not just one. And cartridges really do have some "sense" of CD (the stylus compliance winds up doing a crude highpass). That said, what you are referring to is very specifically a form of rumble, and one that exists at a far lower frequency than what is implied here.

Your graph only goes down to 100Hz.  It's certainly being high-passed.  Take a 1kHz waveform, introduce clipping by amplifying it by 4 dB over full scale, bring the signal back down by attenuating by 6dB.  Then apply an 18th order high-pass Butterworth @20Hz and tell me what you get.

Just a guess...  High-pass filtering???  I don't know what frequency/period we are looking at...  For example, if you high-pass a square wave, you will retain the waveform's vertical step, but the horizontal part will droop-off.

There is probably a high-pass (maybe subsonic) filter somewhere in the cutting-lathe chain, and the record & stylus will also form a mechanical high-pass filter (with a very low cut-off frequency). 

If it is, say a 20Hz high-pass filter, we can duplicate that digitally!

-- EDIT --
Thinking/writing too slow...  Beaten to the punch!

Your graph only goes down to 100Hz.  It's certainly being high-passed.  Take a 1kHz waveform, introduce clipping by amplifying it by 4 dB over full scale, bring the signal back down by attenuating by 6dB.  Then apply an 18th order high-pass Butterworth @20Hz and tell me what you get.

Ask and you shall receive, but it's not doing any favors for the highpass idea:

1s window, 5hz-2khz

20hz!?! No. There is no visual evidence of highpassing at that frequency (or any frequency that low) on waveform plots. There is also no evidence for that on spectral plots (as I've just pointed out). If anything, the highpassing would need to be occurring in the 100-200hz regime to make sense on the plots - and the spectral charts rule that out too.

It's a high-pass, Axon.  No question about it.

[ Specified attachment is not available ]

Greynol, you're a bastard.  I could not believe that a 20hz filter could affect a 1khz-fundamental signal like that. However, you're partially wrong, but you raise a seriously interesting point.

A 18th order Butterworth highpass filter at 20hz has 10 degrees of phase shift at 1khz, 6 degrees at 2khz and 1 degree at 10khz. The phase shifts of the clipped sine wave are what is causing the plot to shift. You are observing an allpass filter - or, at least, you're using a highpass filter as an allpass filter.

An 18th order filter is kind of an extreme case, but the same basic principle should also apply to the second-order filter defined by the tonearm-cartridge resonance - the phase response may be significant enough at midrange/treble frequencies to behave like an allpass. And virtually all turntables have a major tonearm-cartridge resonance. It behaves both as a highpass filter (below the resonance) and as a major bass boost at the resonance, and it allpasses higher frequencies in a non-linear-phase fashion. It could explain everything!

The smoking gun here would be to either a) get a hold of a guy with a fluid-damped tonearm and the DM LP and see if this goes away, or b) implement the tonearm resonance in software and show that it matches the needledrop.

I considered the nature of the tonearm and needle, especially after it was raised by [JAZ].  Either way it is a high-pass response.  You'll see that I never said where or how the high-pass was applied (DVDdoug could very well be right).

Regarding your comment about an all-pass filter, no, it's a high-pass filter.  Regarding the phase response, pick any variety of high-pass you like, you'll still see the clipping slope.

So I cooked up a 2nd-order highpass filter and I've found that I can get roughly the amount of skew needed by setting f0=30hz and the damping factor to 0.3 or so. That's an extremely high resonant frequency and a rather high damping factor but this is quickly getting into the realm of possibility here. No smoking gun yet though (ie a filter which is known to exist in the playback chain which causes the requisite phase shift and matches known frequency response comparisons).

These are CD vs vinyl waveform plots. The CD plots are the ones with the razor-flat horizontal clipping lines.

Don't trust your plots blindly, most applications produce inaccurate waveform representations since they don't use oversampling.
I'd like to suggest that you put your CD audio through a DA/AD chain and look again.

So I cooked up a 2nd-order highpass filter and I've found that I can get roughly the amount of skew needed by setting f0=30hz and the damping factor to 0.3 or so. That's an extremely high resonant frequency and a rather high damping factor but this is quickly getting into the realm of possibility here.

I don't think you need to look to tonearm/cartridge resonances for the source of this. Vinyl records are routinely high-passed as they are cut, and 30Hz seems like a fairly typical corner frequency.

Not being one to gloat, but I gave you the same answers exactly one month ago in the first paragraph of THIS post.

Cheers, Slipstreem. 

I asked at PSW if they've ever heard of using an allpass in mastering, and one vinyl engineer shot that idea down cold.

Some compressors have dynamic all-pass filter-like behaviour. These are certainly used in radio broadcasting - I'm sure similar equipment makes it way into some recoridng/mastering studios.

Don't trust these people to know what their equipment actually does. They're "arty" types. To actually understand the technology would make them too "geeky" to work in their "profession".

This is an extremely important question because some mastering engineers are using these waveform plots as supporting evidence to show that the vinyl masters are different from and superior to the CD masters.

Well, that's just silly. It says...

The upper (CD) waveform is digitally clipped or "squared off", as expected, but the unexpected part is the vinyl version - it's a very unusual, unnatural shape because of the distortion, but it's not squared off. This explains why the vinyl sounds better - although harsh, the analogue distortion hasn't completely removed all the remaining dynamics. Extreme digital clipping of this kind on the other hand, where the whole wave becomes almost square, obliterates pitch information - the ear can't resolve the original fundamental or it's harmonics. Impact and punch are lost, too - the result sounds two-dimensional and plastic in comparison.

Some people on the Metallica forums expressed the opinion that the extra "peak" information on the vinyl didn't represent "real" musical information, simply the inherent difference in the format. Initially, I was inclined to agree, but after doing these listening tests myself I don't agree - the vinyl is less clipped than the CD and sounds better as a result.

A best guess would be that the vinyl is mastered from the CD (or at least the same master) - and that the additional "analogue" distortion on the vinyl version has sweetened or softened the "digital" distortion on the master itself. Whether it's closer to an undistorted master is anyone's guess - probably not. It's an expensive (not to mention silly) way of sweetening the sound. Much easier to buy the CD and throw some DSP at it. You could just record the CD onto tape, and then play it back. Or the record company could simply trash the CD version less to start with!


If they can make a single trashed master, release it on vinyl, charge more for it, and then have fans claim that it's not quite as bad as the CD, they must be laughing!

If this is music you really love, then I suppose you might be driven to all kinds of extremes to try to find/make a better sounding version.

Cheers,
David.

P.S. As others have said, I don't understand how the author can believe the vinyl is from a less compressed master - the graphs show exactly the same level of clipping, plus a high pass filter - that's it!

the graphs show exactly the same level of clipping, plus a high pass filter - that's it!

Plus some low freq crap (mains hum or 2nd harm.?) -- this is what makes it look more dynamic (zoomed out).

  -bandpass

A best guess would be that the vinyl is mastered from the CD (or at least the same master) - and that the additional "analogue" distortion on the vinyl version has sweetened or softened the "digital" distortion on the master itself. Whether it's closer to an undistorted master is anyone's guess - probably not. It's an expensive (not to mention silly) way of sweetening the sound. Much easier to buy the CD and throw some DSP at it. You could just record the CD onto tape, and then play it back. Or the record company could simply trash the CD version less to start with!

I have the feeling 2Bdecided is absolutely right. That could explain my limiter assumption
Anyone can try a maximizer on the already clipped cd and look at the waveforms?

If this is music you really love, then I suppose you might be driven to all kinds of extremes to try to find/make a better sounding version.

In the case of the Metallica album, they already did. The Guitar Hero version is far superior to both CD and vinyl, sound-quality-wise.

So this can be considered as objective and conclusive proof that Guitar Hero as a distribution media has much better sound quality than both CD and vinyl. Finally the ages-old war is over

Classic. I've just listened to the guitar hero version. Just a little mastering, and it would sound great - far better than the CD version.

At the moment, the GH version sounds "unmastered" - IMO it should sound like the CD version, but without the clipping and distortion - then the drums would have the full (i.e. compressed) sound they wanted, but without distortion.

Cheers,
David.

IMO it should sound like the CD version, but without the clipping and distortion - then the drums would have the full (i.e. compressed) sound they wanted, but without distortion.

Tried the D34DL1N3R version? To me it sounds very close to the original just without distortion.

What causes this sloping?

FWIW, I definitely see that kind of sloping in the waveforms of *CD* of Amy Winehouse's hit album (Back to Black).  I presumed it was some form of purposeful production distortion, given that album's aggressive attempt to sound 'retro'. For the sound of that CD, I wouldn't be the least surprised if it was due to 'high pass' filtering.

here's an example from 'Rehab' (image from Audition 1.0). THis is from the 'deluxe' edition of the CD, which supposedly had better mastering than the standard issue.

I don't think you need to look to tonearm/cartridge resonances for the source of this. Vinyl records are routinely high-passed as they are cut, and 30Hz seems like a fairly typical corner frequency.

That's the usual answer - that the elliptic filter is to blame here - and I'd be quite happy if it were that easy. But the spectrum plots don't back that up and at least one mastering engineer (SH, and possibly KG too) has stated that they do not use an elliptic filter with what they cut, so it's not as universal as one may believe. That said, they may be full of sh*t, I don't know.

Some compressors have dynamic all-pass filter-like behaviour. These are certainly used in radio broadcasting - I'm sure similar equipment makes it way into some recoridng/mastering studios.

Right (and well, straight up allpass filters are used in broadcasting too). But I'm having a devil of a time actually proving that it could be in use in these instances.

Don't trust these people to know what their equipment actually does. They're "arty" types. To actually understand the technology would make them too "geeky" to work in their "profession".

Bah! I'm supposed to be the cynical one in this conversation! Not you! 

I more or less agree with you, but Paul Gold at PSW, at least, seems to know basically what he's talking about, and he was the guy who shot down the allpass idea. I mean, you can't be a successful cutting engineer without knowing your signal chain inside and out. That said, Stan Ricker has been happily answering peoples' questions over email, and I have not asked him about this yet....... and he, of course, speaks with the authority of God(dess) on the matter.

A best guess would be that the vinyl is mastered from the CD (or at least the same master) - and that the additional "analogue" distortion on the vinyl version has sweetened or softened the "digital" distortion on the master itself. Whether it's closer to an undistorted master is anyone's guess - probably not. It's an expensive (not to mention silly) way of sweetening the sound. Much easier to buy the CD and throw some DSP at it. You could just record the CD onto tape, and then play it back. Or the record company could simply trash the CD version less to start with!

If they can make a single trashed master, release it on vinyl, charge more for it, and then have fans claim that it's not quite as bad as the CD, they must be laughing!

If this is music you really love, then I suppose you might be driven to all kinds of extremes to try to find/make a better sounding version.

  Precisely my point. Or, at the very least, largely my point.
 
  Ian's still arguing that less clipping = better, which is great. But I think he's mudding the waters significantly by claiming the vinyl version is better without getting to the bottom of why it sounds better. Attributing it to a mix change is a classic audiophile fallacy (attributing audible differences to intrinsic properties rather than extrinsic ones).

Certainly, if he were to conclude that the source mixes were the same, we'd all be in quite a pickle: producers could come out of this whole mess saying "Look, vinyl mastering makes our hypercompressed record sound better. Vinyl mastering is a value add for our product, regardless of the source master. So we'll just source our LPs from our CD masters and charge double for them. And listeners will be getting their money's worth."

Nobody wants to see that happen, so it's a lot nicer to think that the source material really is superior. (I'm not trying to insinuate anything, I'm just trying to state that people have a hard time ascribing sound quality improvement to wholly frivolous effects. And I'm not saying that Ian etal falls into that trap - I'm saying that he may be falling into that trap, and many more people may follow him, and it will require extraordinary proof to get them out of it.)

But the big risk with that conclusion - that the sources differ - is that Ian's analysis is going to apply to virtually all modern vinyl releases - including, I suspect, those which were quite definitively sourced from CD. That is, my fear is, virtually all needledrops will peak higher than the same material on CD, and virtually all of them will have skewed clipping peaks and louder bass due to analog effects rather than mixing changes, and they will always have less objectionable distortion... So those hypercompressed records will continue to be purchased, their owners continuing in the belief that they are buying a higher-quality, less-compressed product. That will make Mikey Fremer and the vinyl industry happy but probably not anybody else.

Stated more simply: That even audio professionals may not be able to accurately tell the provenance of a vinyl master is a very bad sign for those wishing to see mixing/mastering quality improve.

This also suggests that our analysis tools are woefully inadequate for the task, or at the very least, the good ones are not well known enough.... Something that I at least try to correct with my bag o' tricks, like my spectral comparison code. Is anybody interested in seeing it released?

P.S. As others have said, I don't understand how the author can believe the vinyl is from a less compressed master - the graphs show exactly the same level of clipping, plus a high pass filter - that's it!

So, how much should this matter be pursued? Is this all a tempest in a teapot or could "real damage" be inflicted by these statements?

the graphs show exactly the same level of clipping, plus a high pass filter - that's it!

Plus some low freq crap (mains hum or 2nd harm.?) -- this is what makes it look more dynamic (zoomed out).

As my plots indicate, the freq boost is pretty real - it's a roughly +10db 1 octave boost around 25hz, plus another +10db of 45rpm harmonics (which are effectively not "real" and may be ignored).

It's really tempting to chalk up this boost to the arm resonance, and that would be a neat explanation for the phase error, but SRSLY now - when was the last time you ever saw an arm resonate at 25hz? You'd have to put a freakin' ceramic cart on the thing to get that nowadays.

You know, given that this whole debate over the GH3 mixes etc has been 2+ weeks old, I've kind of wondered why HA has had so relatively little discussion on the Death Magnetic phenomenon recently. Has everybody but me been busy at the AES convention or something? 

What causes this sloping?

FWIW, I definitely see that kind of sloping in the waveforms of *CD* of Amy Winehouse's hit album (Back to Black).  I presumed it was some form of purposeful production distortion, given that album's aggressive attempt to sound 'retro'. For the sound of that CD, I wouldn't be the least surprised if it was due to 'high pass' filtering.

here's an example from 'Rehab' (image from Audition 1.0). THis is from the 'deluxe' edition of the CD, which supposedly had better mastering than the standard issue.

I can do ya one better:

http://www.hydrogenaudio.org/forums/index....ost&id=4696

The DM CD, as Ian (correctly!) points out, shows two levels of clipping, in the top plot. The one applied earlier is the sloped section, and the one applied later (in mastering ostensibly) is the flat one.

But Ian (incorrectly!) believes that the later clipping does not exist in the vinyl, shown in the bottom plot. I'd come to the opposite conclusion judging from here.

Regarding your comment about an all-pass filter, no, it's a high-pass filter.  Regarding the phase response, pick any variety of high-pass you like, you'll still see the clipping slope.

Nitpick: No, an FIR highpass would not exhibit this, nor would a bidirectional IIR highpass (filter in one direction then reverse the signal and filter it again).