Digital vs Vinyl

[Newman]

Digital for Factuals, vinyl for Believers.

 

[board]

....and if the end result is that you can compress a song (not in any super-tricky, special way) to get DR score down by 6 points, then struggle to tell them apart / clearly prefer one / hear anything obviously detrimental about the compressed version, then, a 6 point lower DR score is no way to even necessarily be suspicious that the compression itself is detrimental to that version of the song. It could be.... or it could be not.

In which case, the TTDR tool is worse than useless: it is actively deceptive and misleading.

cheers
Grant

This is not meant as criticism, but I find that this topic of the DR meter is turning into a war.
Although I'm not a big fan of the DR meter either, then I think you and Levimax agree a lot more than you might think.
At least he has said that it's a flawed and deceptive tool that however has its uses in certain situations. I find that you then turn it into a bit of a war about what descriptive words to use about the tool, as you seem intent of proving that the tool is, as you say, broken. But I actually think you and Levimax agree more than you disagree. At least I think you also agree that the DR meter shows somewhat meaningful numbers for digital material that hasn't been (overly) dynamically compressed.
I think the disagreement really lies in the following: You notice the problematic aspects with the DR meter (e.g. for vinyl), and therefore you dismiss the use of the tool (almost) completely, although I think you actually do acknowledge that it can show meaningful information in some situations. However, you then say that because of these problematic issues the tool shouldn't be used at all.
Levimax acknowledges the same issues, e.g. with vinyl, but nevertheless says he can choose not to use the tool for those cases or at least not take those numbers to mean anything.
So crudely put, the heart of the matter to me seems to be that you amplify the problems the DR Meter, while Levimax trivalizes them.
I think the point that Levimax was trying to make was that the DR meter is (perhaps) the only tool we have at the moment, and until we find something better we can use the tool at our own peril, whereas you seem to say that it shouldn't be used at all, and we should wait with using such a tool until a proper tool has been developed.
But I still think both you, me and Levimax agree that it's a faulty tool that still can show useful results in some situations - I think the argument really is about how large a percentage of the results are useful, where you say it's very few, while Levimax says it's most. The truth is probably to be found somewhere in between.
I think the DR meter was never developed with vinyl in mind, and then because it can be "tricked" by vinyl it has become deceptive when it fell into "the wrong hands" (i.e. vinyl lovers), which was something the developers hadn't anticipated, and I would imagine that they probably didn't have any idea for how to code the program to avoid it being "tricked" by vinyl.

 

[Newman]

Hi board,

the DR meter shows somewhat meaningful numbers for digital material that hasn't been (overly) dynamically compressed.

Well, if the material hasn’t been overly dynamically compressed, it isn’t part of the audiophile use of the tool, which is primarily to identify material that has been overly dynamically compressed.

What audiophiles need the tool to do well, to serve their primary purpose, is to show somewhat meaningful (reliable, consistent) numbers for digital material that has been (overly) dynamically compressed.


-

How do you know exactly when it is reliable and when it isn’t? Have you tested it in every possible situation, i.e. every type of dynamic compression, plus every type of mixing/mastering edit other than dynamic compression, to see which ones inadvertently cause the DR score to change when it shouldn’t? And, in doing so, what were you using as your reference tool to check TTDR against? @levimax wants to know: he says there is no other. OTOH if you don’t have one, how have you come to your cautiously optimistic conclusions?

How do we know the full suite of its failings? I am aware of the digital vs vinyl failing; the blending the bass failing; the 6-point-drop-with-compression-but-no-clearly-audible-detriment failing. I am aware of the corrosive, confirmation bias effect, where people are starting to ‘hear’ what the tool is saying, even when they are wrong (eg both versions produced from the same master but different TTDR scores). To me, that confirmation bias issue is a big failing, even though it’s actually a flow-on effect. It still only happened because of the tool’s failings.

How much is enough?

I think the DR meter was never developed with vinyl in mind, and then because it can be "tricked" by vinyl it has become deceptive when it fell into "the wrong hands" (i.e. vinyl lovers), which was something the developers hadn't anticipated,

And here we have some common ground, albeit not just for vinyl: other all-digital failings have been established. But I firmly agree that the developers made it for use other than how it is being used by audiophiles. That is easy to agree on, because the developers made the plug-in for studio engineers to use during mixing and mastering. And I don’t have any issues or any problem with that whatsoever. Whether or not there is a bit of inaccuracy here or there, sure whatever, the studio engineer has complex objectives in mind and the TTDR plug-in might be useful sometimes, so go for it. I’m not referring to it as broken or inappropriate for that application whatsoever.

Here is what the developers intended in their own words: “The MAAT DRMeter is a worthy companion for daily mixing and mastering tasks, enabling fast, accurate and easy mix and mastering decisions. For fine tuning peak excursions, and properly setting your limiter with the fast and responsive peak meter bar, DRMeter is unmatched.”

Like you say, the developers had not anticipated their plug-in tool for studio engineers being used in the way it is by audiophiles: by consumers at the end point to judge the final retail product on a master versus master basis. That is the application for which I say it is effectively broken or inappropriate, and the reason for that is it there are numerous ways that it is not measuring dynamic range, that is varying with other variables, that is deceiving people into thinking they are hearing the numbers that it says when they aren’t, that people are making purchase decisions on the basis of this, or recommending to each other not to buy something or to buy something on the basis of this number (always with the confirmation bias that they hear it as reported by the tool). In other words, the developers intend it to be used where only one variable is being changed, to assist in metering how much of that variable has been changed for scale and consistency: but audiophiles are using it on an end product where every variable has been changed, assuming it will isolate how much just one variable has been changed. That seems to be not what it is made for, and there is evidence that is not what it can do — not with the reliability and consistency that would make it useful for that purpose. But audiophiles are assuming it can and does.

Cheers

 

[danadam]

Here is what the developers intended in their own words: “The MAAT DRMeter

Is MAAT DRMeter the same thing as the very simple original algorithm? AFAIK it is not. Does it suffer the same limitations mentioned in this thread? (Serious question, because I don't know).

 

[mhardy6647]

Is there a TLDR of who won?

Puts me to mind of Wooden Ships.

If you smile at me, I will understand
'Cause that is something everybody everywhere does in the same language
I can see by your coat, my friend, you're from the other side
There's just one thing I got to know
Can you tell me please, who won?
...

 

[board]

Well, if the material hasn’t been overly dynamically compressed, it isn’t part of the audiophile use of the tool, which is primarily to identify material that has been overly dynamically compressed

Reading your response I see that what I wrote apparently wasn't as clear or precise, or maybe even correct, as I had hoped.
I have singled out a single piece of your post. This is exactly the point that Levimax was making: He doesn't use the DR meter (abuse it) for comparing vinyl to CD - he uses it to distinguish between digital material that has been dynamically compressed and material that hasn't (and yes, I know that all music has been at least compressed with an analogue compressor, but I think you know what I mean). And I think this is where you and him, and me, find common ground.
Wouldn't you say that the DR meter is at least somewhat useful for distinguishing between those two types of material?
It may be useless for other uses, but that's a different topic.
I'm asking that question, because I think your answer is yes, but it's possible that I have misunderstood you .

 

[Newman]

@board Thanks for taking the time to read my somewhat rambling post.

to distinguish between digital material that has been dynamically compressed and material that hasn't…Wouldn't you say that the DR meter is at least somewhat useful for distinguishing between those two types of material?

It should be, as long as it doesn’t do any of the following:

• Jump up or down by anything up to 6 points with an action unrelated to dynamics, e.g. blending the L&R bass. (Ian Shepherd demo)
• Deliver a 6-point drop in DR score when the amount of compression applied results in an inaudible / barely detectable difference. (Your demo)

Unfortunately, it does do both of those. And both the above cases were not edge cases: they were routine things.

So…..how can we use it to draw any conclusions about 2 commercial final releases that we don’t already own, but see the DR scores on the loudness wars database?

 

[board]

@board Thanks for taking the time to read my somewhat rambling post.



It should be, as long as it doesn’t do any of the following:

• Jump up or down by anything up to 6 points with an action unrelated to dynamics, e.g. blending the L&R bass. (Ian Shepherd demo)
• Provide an inaudible / barely detectable difference, if enough compression is added to produce a 6-point drop in DR. (Your demo)

Unfortunately, it does do both of those. And both the above cases were not edge cases: they were routine things.

So…..how can we use it to draw any conclusions about 2 commercial final products that we don’t already own, but see the DR scores on the loudness wars database?

I actually originally wrote about the summing of low frequencies or the use of a high-pass filter in my last post, but then decided to delete that part before posting. I wrote that it wasn't relevant, and it actually isn't:

As mentioned, the scenario would only be with digital material, so not material cut to vinyl, and summing of low frequencies and/or the use of a high-pass filter would be something you would only use for cutting vinyl. So it might be routine for cutting vinyl, but not with creating digital material. Those two processes are only used after the digital master has been finished and that material then has to be cut to vinyl, and those two processes are used due to physical restraints with vinyl that are not present with digital. As mentioned, we already accepted that the DR meter is not useful for use with vinyl for that particular reason.

About producing an inaudible/barely detectable difference by using heavy compression, it looks to me like you're making the opposite point of what you think you're making. The point of using dynamic range compression is exactly to make it louder (dropping the DR by 6 points as you say), while at the same time making the effect inaudible/barely audible if you lower the volume to match the uncompressed material, but obviously it's completely audible if you don't lower the volume (i.e. it will be louder, which is the point of dynamic range compression).
So, in that way the DR meter works exactly like it's supposed to for that purpose, and it only really becomes very unreliable when using it for vinyl.

I do dislike the DR meter for other reasons though, but that is mainly because in some cases (but far from always) it measures one song higher than another, even on the same album, even though the RMS value is higher on the song that measures the lowest in DR (e.g. one track: DR7. RMS: -8.21 dB. Another track from the same album: DR5. RMS: -9.69 dB. Another album: One song: DR12. Peak: 0.00 dB. RMS: -14.97 dB. Another song on the same album: DR10. Peak: 0.00 dB. RMS: -14.97 dB. And yes, those were actual tracks I measured). So please don't think I'm singing the praise for the DR meter, but for the specific purpose of comparing dynamically compressed digital material to uncompressed digital material it is at least somewhat useful. To sum it up: It's better than nothing. Or maybe more precisely: It's all we have, but here's hoping for an improvement in the future.

 

[Newman]

the scenario would only be with digital material, so not material cut to vinyl, and summing of low frequencies and/or the use of a high-pass filter would be something you would only use for cutting vinyl. So it might be routine for cutting vinyl, but not with creating digital material.

Sure, but deciding on the original L/R mix of bass tracks applies to digital too. If summing L/R bass for vinyl causes DR score to swing, so will the original mix decision.

About producing an inaudible/barely detectable difference by using heavy compression, it looks to me like you're making the opposite point of what you think you're making. The point of using dynamic range compression is exactly to make it louder (dropping the DR by 6 points as you say), while at the same time making the effect inaudible/barely audible if you lower the volume to match the uncompressed material, but obviously it's completely audible if you don't lower the volume (i.e. it will be louder, which is the point of dynamic range compression).
So, in that way the DR meter works exactly like it's supposed to for that purpose

Sure, but if you have compressed by 6 dB and it is inaudible / barely audible (other than loudness), then people shouldn’t be looking at the scores and saying, “oh, that one with the low score is undesirable because it will sound terribly compressed by comparison.” But they do. That’s exactly how they use the tool. Then they get confirmation bias.

And that’s the problem. And it applies to digital vs digital.

cheers

 

[board]

Sure, but deciding on the original L/R mix of bass tracks applies to digital too. If summing L/R bass for vinyl causes DR score to swing, so will the original mix decision.

Yes, that but's not relevant, because we're not talking about how to mix. Also, the original mix is on the CD, whether it's the original master or the remaster. In a remaster the mix hasn't been changed - then it would be a remix.
On a slightly different note, I don't think bass is being panned nowadays anyway. That seemed to be a 1960s trend.

Sure, but if you have compressed by 6 dB and it is inaudible / barely audible (other than loudness), then people shouldn’t be looking at the scores and saying, “oh, that one with the low score is undesirable because it will sound terribly compressed by comparison.” But they do. That’s exactly how they use the tool. Then they get confirmation bias.

And that’s the problem. And it applies to digital vs digital.

cheers

That's a different topic. You're now talking about how people interpret the numbers the DR meter show, not if the tool actually works. Then your mission shouldn't be to discredit the tool, but rather to educate the people using it.

 

[Mart68]

Sure, but deciding on the original L/R mix of bass tracks applies to digital too. If summing L/R bass for vinyl causes DR score to swing, so will the original mix decision.



Sure, but if you have compressed by 6 dB and it is inaudible / barely audible (other than loudness), then people shouldn’t be looking at the scores and saying, “oh, that one with the low score is undesirable because it will sound terribly compressed by comparison.” But they do. That’s exactly how they use the tool. Then they get confirmation bias.

And that’s the problem. And it applies to digital vs digital.

cheers

regardless of that. if you buy the version with the high DR score you can't go wrong. Maybe you could have bought the version with the low DR score and in reality its DR would be the same as the version with the high score.

But the relevant objective has still been achieved - you have bought a high DR version.

 

[rdenney]

TL;DR: All models are false, but some are useful.

DR isn't a tool, it's a model. Some use models as tools, and make what is supposed to be descriptive into an objective. As I have said many times in my own field, if we lack objectives, our performance measures (models) will become our objectives. Objectives should come first.

In the vinyl period, compression was used to keep the needle in the groove. And do sound reasonable on Decca record players. That was the objective.

In the FM Radio period, compression (really, modulation limiting--such as by use of the Optimod) was used to make the signal audible in a noisy car, even the quiet bits. This was really the Loudness War writ large. It has not gotten better since then, either. (Robin might argue that it is necessary, but if I need to gain-ride a classical recording to hear the pp oboe solo in between the loud orchestra tutti sections when listening in the car, I know where my volume knob is.)

In the portable cassette player period, recording often used automatic gain control--a form of compression--because casual recordists were unable to set levels properly. But it meant that people knew what gain-riding sounded like, even when it was done well (and didn't cause background hiss to pump).

All those periods overlapped, of course, and so we had different objectives that also sometimes overlapped.

I don't look at DR scores, but I do listen to recordings. I find that the first CD's often have more "punch" (less compression) than the vinyl records that were compressed from the same masters. I find that digital remasters of those albums often clean off the background hiss, but also lose some of the clarity that I associate with a lack of compression. Examples: Rick Wakeman's Six Wives of Henry VIII in vinyl (compressed) versus the first CD from the same masters (less compressed, but subtly and may be the difference between my CD player and my turntable). Another example: Rick Wakeman's Criminal Record in vinyl (less compressed) versus later Japanese CD pressing (more compressed and therefore muddier). I've not made measurements, but I have made comparisons. So, my own model is quite imprecise, but I don't confuse it with objectives.

Rick "cringing when seeing reports of producers using DR as an objective for 'optimization'" Denney

 

[Robin L]

(Robin might argue that it is necessary, but if I need to gain-ride a classical recording to hear the pp oboe solo in between the loud orchestra tutti sections when listening in the car, I know where my volume knob is.)

First, no one should be distracted while driving. Second, it is necessary. One doesn't want overmodulation, and one doesn't want to lose the reach of the signal. So even the good FM stations, like the one I volunteered at, the nation's first listener sponsored station, used the Optimod. I could always hear the pumping on my recordings over the air because my digital recordings were not limited in post production before broadcast. I did use reverb, sometimes EQ, would "pencil" out the more egregious peaks [orchestra music's like that], but never used a compressor or limiter.

 

[rdenney]

First, no one should be distracted while driving. Second, it is necessary. One doesn't want overmodulation, and one doesn't want to lose the reach of the signal. So even the good FM stations, like the one I volunteered at, the nation's first listener sponsored station, used the Optimod. I could always hear the pumping on my recordings over the air because my digital recordings were not limited in post production before broadcast. I did use reverb, sometimes EQ, would "pencil" out the more egregious peaks [orchestra music's like that], but never used a compressor or limiter.

Distracted while driving?

Well, then, I can't listen to Shostakovich, or Vaughan Williams, or... they all command attention.

But the volume control? That's two buttons on my steering wheel! I don't even have to take my hands off the wheel to adjust the volume.

Rick "who still listens to classical music over the air--at home--and hears the pumping" Denney

 

[mhardy6647]

TL;DR: All models are false, but some are useful.

 

[DVDdoug]

[QUTOE]Compression is nothing more or less than added distortion... [/QUOTE]You can say it's a kind of distortion (dynamic distortion) but when done properly it doesn't sound like distortion and it doesn't always measure as distortion. You could say also say that EQ or reverb are distortion because they alter the signal, but these are also part of everyday audio production.

It does reduce the signal-to-noise ratio and since it's often used to make "everything louder" the existing noise and distortion are boosted.

On the other hand, clipping is harmonic distortion and the worst kind of dynamic compression. A traditional limiter (without look-ahead) can distort the waveform.

 

[Robin L]

Distracted while driving?

Well, then, I can't listen to Shostakovich, or Vaughan Williams, or... they all command attention.

But the volume control? That's two buttons on my steering wheel! I don't even have to take my hands off the wheel to adjust the volume.

Rick "who still listens to classical music over the air--at home--and hears the pumping" Denney

But you're in minority on multiple counts. Again, the real concern is for reach, the ability to be clearly audible in the outer reaches of an FCC approved broadcast area. And who's broadcasting Shostakovich these days anyway?

Robin "who stopped listening to FM years ago" L

 

[egellings]

Part of vinyl's draw is the fiddly pet-like nature of it and the TTs, arms & cartridges it's played on. Oh, and it can sound quite good. With a digital source, you have this shiny object to stare at. And it sounds perfect. I have both kinds of source. The digital one is more capable than the vinyl one is, what with all the dynamic range, low distortion & noise and all.

 

[rdenney]

But you're in minority on multiple counts. Again, the real concern is for reach, the ability to be clearly audible in the outer reaches of an FCC approved broadcast area. And who's broadcasting Shostakovich these days anyway?

Robin "who stopped listening to FM years ago" L

If you stopped listening to FM years ago, then how would you know who is or isn't broadcasting Shostakovich?

But I listen to a thumb drive in my car, and I'm happy to gain-ride if necessary. But it isn't really necessary, because my car is fairly quiet.

WETA, by the way. They are playing (just the example off the top of my head) Lady Macbeth of Mtsensk in a couple of weeks. But, again, I usually listen to them at home, in the evenings, when I'm doing other things and just want a stream of stuff I might not have heard before.

I built this antenna primarily to receive WETA (also over-the-air television to improve WAF):

Rick "yes, on the fringe--in terms of reception, assuredly" Denney

 

[mhardy6647]

If you stopped listening to FM years ago, then how would you know who is or isn't broadcasting Shostakovich?

But I listen to a thumb drive in my car, and I'm happy to gain-ride if necessary. But it isn't really necessary, because my car is fairly quiet.

WETA, by the way. They are playing (just the example off the top of my head) Lady Macbeth of Mtsensk in a couple of weeks. But, again, I usually listen to them at home, in the evenings, when I'm doing other things and just want a stream of stuff I might not have heard before.

I built this antenna primarily to receive WETA (also over-the-air television to improve WAF):

Rick "yes, on the fringe--in terms of reception, assuredly" Denney

Is that a Toyota T100 I espy or "merely" a Taco? Either way -- nice truck