MQA is Vaporware

[Jud]

Just now, mansr said:

 When have either of those discussed compression algorithms? Their products do not even use them.

 

Mike Moffat very definitely has discussed the term "lossy," in the sense of mathematical irreversibility, with regard to sample rate conversion and the accompanying filtering.  Miska has of course discussed his filters, closed form and otherwise, in the context of reversibility/irreversibility, but I am not certain whether he ever referenced those conversions as "lossy" or "lossless."

 

Just now, mansr said:

I think there's a chance that it might for some inputs. Since I haven't listened to every available MQA track, I can't say whether it has happened yet.

 

I'll "jump the gun" rather than waiting for you to listen to every available track and ask, of those you have listened to, do you think it has degraded sound quality?

[Jud]

2 minutes ago, crenca said:

 

 

Jud, you have said this before - however the modulations going on inside ADC & DAC's is besides the point.  The point being the definition of a file, a format, a software encoded (or for that matter an analogue "encoded" format such as vinyl) piece of software that is a static "capture" of the signal/impulse.  "lossy" has definition in this domain, and it is this domain that is under discussion and important.  Why?  Because it is the "stuff" of what every DAC has to work with, every consumer has to work with.  It is also the point of attack by Bob's/Meridian's lies and it is where consumers should be aware of the obfuscation going on...

 

 

"Lossy" and "lossless" have only one precise definition, a mathematical one.  "Lossless" = mathematically reversible, "lossy" = mathematically irreversible.  I agree this definition is used by far most often in the context of compression.  However, I've read at least one engineer who was working on digital audio filtering before some of us were born use those terms to refer to the filtering accompanying sample rate conversion.  Certainly the mathematical definition is exactly the same.  The only question is whether it is appropriate to use the term when talking about something other than compression.

 

And of course this is only interesting if we want to quibble about definitions.  I'm not sure why that's necessary, since I'm agreeing with you and mansr that MQA unnecessarily uses lossy compression (and even further, that the "unnecessary" part distinguishes it from such formats as MP3 and AAC).  Perhaps mansr wants to be able to apply the term "lossy" exclusively to MQA because it has pejorative connotations and so far in his listening he hasn't noticed a sound quality difference?

[Jud]

11 minutes ago, crenca said:

 

Not the usual/widely used definition and (this is the important point) NOT the way Bob/Meridian mean when they misuse the term for their own purposes...but of course you already know that.

 

You will want to pay attention to what I am saying.  Once again, I repeat that MQA uses lossy compression.  This is the precise mathematical definition, and exactly the *opposite of* the one Bob/Meridian/MQA are using when they say they are using lossless compression in the sense of "audibly lossless."  This is touchy-feely stuff and not the precise mathematical definition.  Any disagreement?

[Jud]

Just now, mansr said:

The terms can certainly be applied without confusion to any mathematical transformation. However, they make little sense in the context of an operation that is by definition irreversible (e.g. bandwidth reduction).

 

 

Would bandwidth reduction using a closed form filter be impossible?  If not, would it be reversible?

[Jud]

4 minutes ago, crenca said:

 

 

Ok, so you have forgiven his sin of attempting to pull a fast one and then reversing course, and modifying or "clarifying" what he meant...the damage is done however and consumers are "confused" which was no doubt his intention because he is an "audio savant" and knew perfectly well what he was doing...

 

Whatever are you talking about?  What have I forgiven?

 

My discussions with regard to MQA and the term "lossy" have been consistent.  My subjective feelings about what the MQA process does to music have also been consistent.  (I've written here several times that I subjectively feel MQA files sound a little less good to me than their non-MQA 44.1 or hi res counterparts.)

 

So who exactly are you talking to, me or some construct you have in your mind of someone who supports MQA (which I don't)?

[Jud]

26 minutes ago, mansr said:

The terms can certainly be applied without confusion to any mathematical transformation. However, they make little sense in the context of an operation that is by definition irreversible (e.g. bandwidth reduction).

 

 

24 minutes ago, Jud said:

 

Would bandwidth reduction using a closed form filter be impossible?  If not, would it be reversible?

 

21 minutes ago, mansr said:

I don't disagree with that. What Bob calls "lossless," everybody else in the industry has always called "psychoacoustically transparent."

 

Now your agreeing that MQA is lossy does not explain why you're trying to expand the definition of "lossy" to include pretty much everything.

 

Not "pretty much everything," only, as you agree above, mathematical transformations to which the terms "lossy" and "lossless" can be applied without confusion and sensibly (in which regard I anticipate your answers regarding bandwidth reduction via closed form filtering - if it is possible and indeed reversible I'll continue to use the terms; if not, I won't, though I will continue to use them in the context of interpolation, where conversions using closed form filters are indeed reversible, with other filters not reversible).

[Jud]

1 minute ago, mansr said:

Bandwidth reduction means cutting off the high-frequency content. Once that has been done, it is gone forever no matter what kind of filter was used. It can no more be restored than can colour to a black-and-white photograph.

 

Then I'll restrict my use of the terms to interpolation.

[Jud]

Just now, Sal1950 said:

Aren't you one of the "everything matters" crowd here.

 

And in fact my subjective feeling, if you read a few comments above this (or elsewhere on the forum where I've said it several times), is that MQA files don't sound as good to me as their non-MQA counterparts.  (This is aside from cases where the masters are different.  Better masters are better sounding to me, regardless of whether or not they're MQA.)  But I was asking mansr. 

[Jud]

Just now, Sal1950 said:

And I was winking.

 

 

And I was smiling. 

[Jud]

54 minutes ago, crenca said:

 

Have you forgotten that Bob and company changed their tune when others questioned there misuse of the term "lossy" in their marketing materials?  Perhaps so - whatever I was referring back to that...

 

I have no idea what MQA did in its marketing materials, never having read any of them except for one technically oriented white paper.

[Jud]

1 minute ago, Don Hills said:

 

 

This is another example of misdirection, akin to MQA's marketing. What has really happened is that the patents for MP3 have expired, so you no longer need to purchase a license. Instead, Fraunhofer would like you to purchase a license for AAC, for which the patents are still current.  So far from being "the death of MP3", it heralds a new beginning...

 

No no, it is death!! like MQA!!! Ggaahhh!!!

[Jud]

1 hour ago, mansr said:

As I have tried repeatedly to explain, the losses from sample rate reduction (bandwidth limiting) are of a fundamentally different nature than those from a perceptual coding algorithm. They cannot be compared.

 

True, there's probably less research on the audible effects of sample rate reduction.

[Jud]

2 minutes ago, mansr said:

The research says as long as frequencies up to 20 kHz are intact, the change is imperceptible. But you already knew that.

 

And here I thought there was research on time smearing, linear phase vs. minimum phase, apodizing filters, the effect of ultrasonics, etc.  No?

[Jud]

6 hours ago, mansr said:

Is there more than anecdotal evidence that any of those things are perceptible?

 

I believe there is actual research, yes.  But life being busy at the moment, perhaps we could reach a resolution faster this way:

 

@Miska has said he can perceive differences in the sounds of his filters, which vary from each other in the sorts of characteristics I mentioned.  Do you believe him?  If so, these things are indeed perceptible.  Accuracy of those perceptions will then be a matter of training.

[Jud]

40 minutes ago, Jud said:

 

I believe there is actual research, yes.  But life being busy at the moment, perhaps we could reach a resolution faster this way:

 

@Miska has said he can perceive differences in the sounds of his filters, which vary from each other in the sorts of characteristics I mentioned.  Do you believe him?  If so, these things are indeed perceptible.  Accuracy of those perceptions will then be a matter of training.

 

I've just realized I may have put mansr in an awkward personal position by what could be construed as asking him to comment on the statements or work of another developer.  So I withdraw the question.

[Jud]

17 minutes ago, witchdoctor said:

The goal is to increase paid subscribers vs prirates, I see no harm in that.

 

The trouble is that the labels have historically seen customers as pirates or potential pirates.  Apple came along and gave people the chance to preview a song and see if they liked it, and if they did, to purchase it for 99 cents.  This was in contrast to the labels, who wanted everyone to buy $15-$16 CDs on the basis of advertising and hearing one of a dozen or more songs on the radio.  People didn't bother much to pirate what they could get so easily from Apple.  Problem solved!

 

But the industry is still blaming many of its ills on piracy - better than blaming them on the record company execs, eh?

[Jud]

Thanks mansr for the really informative comment.

 

Have you ever noticed that people in this forum (perhaps elsewhere in your life as well?) tend to react much more favorably to your less strident comments (for example, the nice bulleted list of MQA dislikes, refreshingly free of nastiness and exaggeration, you put up earlier in the thread)?  I.e., you're (paradoxically?) more persuasive when you're reasonable rather than railing?  As someone of a scientific bent, perhaps these data should be an indication to you of the road to future success.

[Jud]

1 hour ago, Don Hills said:

 

Unlike the majors, Linn cares about audio quality.

 

So does the 2L label, though.  (Again, I personally think MQA sounds slightly worse than equivalent non-MQA files.)

[Jud]

6 hours ago, esldude said:

 

Do we need a legal copyright/DRM sub-forum and anyone not an attorney can only ask questions of members that are attorneys?

 

Do we need attorneys who are foolish enough to try to answer?

[Jud]

13 hours ago, mansr said:

Do you also want a "balanced" discussion about climate change or evolution vs creationism? You are more and more, as facts about MQA emerge, resembling the science deniers on those topics.

 

12 hours ago, esldude said:

The thing that bothers me besides possible DRM like characteristics of MQA is what it really is.

 

The name was Master Quality Authenticated.  It was supposed to tell us when we are getting the actual bits fully decoded to best possible effect under the MQA system.  Instead you have variations and layers of decoding, and filter adjustment.  Though there are two possible indicator lights that is nowhere near enough to tell the whole story.  Quite often I suspect you will simply get an MQA encoded file that other than being wrapped in that container is little different in quality or provenance.  So on the face of it MQA mostly fails to live up to its name.  All that is left is some file size reduction, and the fact if you don't decode at some level your undecoded file is a slightly adulterated version of a redbook quality file. That is not something I see as a step forward. 

 

Stewart et al have been slow to dribble out details of what is really being done in MQA and are at fault for the rocky reception it has had in some circles.  Promises were made that seemed unlikely and in fact turned out not to be so just as suspected.  Something of a minor bait and switch with a promise to just trust us it will be good.

 

An example of a pejorative comment that doesn't provide information about MQA, and an unpejorative critical comment that does.  (And yes, my own comment here provides no new information.  )

[Jud]

1 hour ago, Melvin said:

Sal .. do you honestly believe the audiophile community can stop this MQA juggernaut? I do not.

 

I think it is far, far from a juggernaut, and yes, I think we can help to stop it by getting good, reliable information out to the press and the majority of prospective purchasers.  Everybody wants something that's better; no one wants to be duped.  I think many people tend to (at least I do) tune out something that comes across as shouting, as some sort of impassioned screed with an agenda.  Give me the facts (@mansr and @Miska's work on what the filtering actually does, and the non-necessity for the lossy compression) in a way that doesn't have to be technical (as @esldude does in his recent comment) and trust me to make a good decision.

 

Edit: Oh and yes, the DRM potential (please not in a way that tries to convince me the barbarians have already broken down the gates, but lets me know the trebuchets are designed and can be built in an instant if need be).

[Jud]

13 minutes ago, Melvin said:

 

Well, the content partners listed on the MQA website suggests differently IMO.

 

 

Let's see how it sells.  (And inform the purchasing public.)

[Jud]

2 hours ago, mansr said:

When we have the original master, there is rarely any reason for us to downsample it at all.

 

But we so rarely have "the original master" in today's market.  We so much more often have Redbook or even mp3/AAC.

 

I dunno, you think it would keep people from spending their hard-earned money if we accurately said this (MQA) is an attempt at a "better" version of mp3/AAC (i.e., better perceptual encoding/lossy compression), but is perhaps not even as good quality as CD (for the same master)?

[Jud]

22 minutes ago, PeterSt said:

Audirvana can decode MQA, right ? Many love the filtering of iZotope, correct ? Now try to combine both.

Or did you already notice that once MQA is decoded that iZotope is not active ?

 

Interesting, will take a look.

[Jud]

31 minutes ago, Albrecht said:

This is simply a brilliant post. Thank you!!

 

And he wrote it himself!