MQA Deep Dive - I published music on tidal to test MQA

[amirm]

Amir, when you setup payments, make them available to non-members as well? I am sure folks at PFM may want to contribute.

Sure but first step is getting someone in the industry that can donate proper music content for this and a willing post house to do the encoding.

 

[blackmetalboon]

What else do we call him? It is not like he has given us his real name.

Where have I heard that before? RMAF 2018?

 

[Daverich4]

The outfit he used doesn't do custom encodings of MQA. For best effort, we need to go to a mastering house that does. Do we have any members in the Pro community that could facilitate providing the content and getting it converted by such a mastering house? We could cough up some money for the service if people value it enough. I will donate up to $250 toward it.

I’d be really interested in this if you’re involved in it enough to have it be on the up and up. If you are, I’d kick in $1,000.

 

[amirm]

Sure seems to imply that they are doing a lossless encoding here by "restoring the original sound" and "unwrapping this perfectly" and "a completely reversibly, lossless process," etc.

I don't understand this criticism. It is not their problem that people can't follow the discussion and generalize to lay understanding. Let me give an example.

The front end of a lossy codec like MP3 is indeed, lossy. It will transform the music to frequency domain, apply psychoacoustics, and generate a new bit stream. This bit stream still has redundancy in it so it is then fed to the back-end that is a totally lossless encoder. It uses a set of code books to assign smaller number of bits to common values, and longer ones to less common ones. So we get further reduction in file size. This can for example deal with silence in a file. And it is completely reversible when the decoder gets the bits.

You can't take what I just explained to say, "oh Amir just said MP3 is lossless." It would be the fault of the non-technical reader to do that, not mine trying to explain how the algorithm works.

And really, in this context, in this forum where we know what the terms mean, I don't understand why this is constantly brought up. If you now know what they mean, we are done with the topic. Constantly bring this up over and over accomplishes what?

 

[DimitryZ]

So I admit I don't know much about the whole MQA situation, which is why I sort of went in with fresh eyes here as an outsider and transcribed those conversations from earlier, to get a more clear outlook of what it is MQA is actually claiming and what value they are providing, if any.

To my understanding, MQA is claiming that their transformation is "better than lossless," arguing that the data contained in lossless is wasteful to begin with, and that the data gets tarnished anyway from time-based "blurring" errors as a result of the analog-to-digital-back-to-analog conversions throughout the whole chain between what you hear in the studio and what you hear on your end as a consumer. And this de-blurring process is done on "the music" (as opposed to the noise), which is identified statistically from what they define as "natural" sounds.

The end result is a smaller filesize containing "relevant" data that has this "de-blurring correction," and presumably sounds much more "natural" with the intent being that you get "what was heard in the studio." This is why MQA tries to insert itself at every step of the chain here, to have that control over how the content is transformed and delivered, with the little light at the very end telling you that the whole process worked and you're hearing "what the artist intended."

So my concerns would be as follows:

1. MQA is not lossless, but they seemingly try to use the term anyway to mislead people into thinking that it's one of their advantages, using careful wording to give themselves plausible deniability if questioned.

There is of course the "music origami" video I posted earlier:



Sure seems to imply that they are doing a lossless encoding here by "restoring the original sound" and "unwrapping this perfectly" and "a completely reversibly, lossless process," etc.



An initial reading makes it sound like "you get the lossless file, which is what the rights owner intended," but technically it's saying that MQA gets it, which says nothing about what they do with the file after the fact.



This one is mad shady to me. Again, an initial reading makes it seem like they're claiming MQA is lossless because it "retains all the detail of the studio recording," which is not technically the same thing as "retaining 100% of the data" if they're implicitly assuming that "detail" is not the same as "data." And "MQA retains 100% of the original recording" implies a different context than "an MP3 keeps just 10% of the data" and yet they're put in contrast within the same sentence despite being an apples-and-oranges thing. They're saying they retain the original recording, and by-the-way MP3 only retains 10% of the data. But again, it's worded carefully to give the impression that they're giving you lossless, making you think they are giving you 100% of the data without specifically saying it.



More careful wording, saying they never technically made any false claims about losslessness (but this is different than making technically-true-but-misleading claims). That said, even on their previous FAQ page, they had "Yes, MQA is lossless," so I'd argue that they technically have made false claims in the past (which would explain the recent edits to "better than lossless"). On that page they try to justify what they mean by this:



Just because I put on a batsuit doesn't make me Batman. Putting lossy data inside a FLAC container and then calling it lossless is really misleading to me.



FLAC is the name of a lossless compression codec, and FLAC also happens to be the name of the container. I mean, it's literally in the name: FLAC = "Free Lossless Audio Codec." Just because you store data in a FLAC container doesn't mean you get to call your data "lossless" compared to the original.



"Delivered losslessly" to me sounds like they're just trying to rely on the ambiguity of "delivery" in context. Are they talking about internet protocols and file transfer integrity? Because that's a different claim entirely than whether the content itself is lossless. Technically this sentence is saying "the file is delivered to you in a format that has been approved." That's it. It's not technically saying the file itself is lossless - but it sure tries to come off that way, in my opinion.

-----------------------------------

Anyway, all of these statements are enough to make me distrust MQA from the getgo. To my eye, they are frequently going out of their way to make carefully-worded, misleading statements to make you think they're saying one thing when ah-ah-ah-they're-not-saying-that. In the interview I partially transcribed earlier, Bob Stuart expresses some dismay at how people are confused about MQA and what it does, but I am not at all surprised when they seem to be causing this confusion intentionally.

2. What's up with ringing / blurring exactly?

If I am understanding them correctly, MQA says that due to the whole analog-to-digital-then-back-to-analog conversions throughout the chain, certain "ringing" and "blurring" effects get introduced. I have never heard of this being some kind of issue that everyone has been scratching their heads over for decades.

The technical guys will be able to say more on this of course since I know nothing about it, but I'd want to see exactly how big of a problem this is to begin with, and be able to transparently compare files / listening experiences in order to see what value MQA's transformations are providing.

I would have assumed that this "issue" (if it exists in any meaningful way) is handled in any quality DAC that tries to faithfully recreate the original analog sound from the digital samples, and I do know that there exists the whole Shannon-Nyquist theorem and all, so I was under the impression that this was a done deal already. I don't claim to know what I'm talking about here, but I think one of the points GoldenSound was making was that MQA makes it difficult to make such a comparison.

From what I can tell, all in all, MQA is offering a lossy compression with their own transformations applied to the sound, getting the rights owner to approve it, and then delivering this content as the "master" under the assumption that "this is was was heard in the studio according to the artist."

But this is not the same as "you're getting the original" or "you're getting the lossless file."

For example, Neil Young pulled out of Tidal, saying as much: https://neilyoungarchives.com/news/1/article?id=Tidal-Misleading-Listeners





3. "If you don't like it, don't pay for it."

That's all fine and well for the individual, but I guess my concern would be more systemic, if enough demand is drummed up over misleading claims and questionable benefits that allow MQA to gain a large amount of penetration and market control, leading to equipment costs going up or even just perpetuating more possible snake oil in an industry that already has plenty of it. In general I am already distrustful of MQA simply due to the way they throw around the word "lossless."

Anyways, would appreciate any responses to these specific points, any clarifications to errors you think I am making here, any elaborations on MQA claims, all that jazz. I'm just an outsider here offering my initial impressions after looking at some MQA content. I did have a subscription to Tidal for a while, but don't any longer because I wasn't a fan of the UI, couldn't find most of the stuff I was looking for, and didn't find there to be any significant benefit to MQA over other formats anyway.

Thanks.

1. Has been done to death and back. Amir has explained it numerous times. I have posted a simple excercise demonstrating it. I even posted a gardening example illustrating this.

It appears that for most people the concepts of data space and musically relevant information are impossible to understand. And if you get stuck here, there is no path forward - what MQA does will forever seem suspicious and LOSSY.

 

[amirm]

Where have I heard that before? RMAF 2018?

You have no case with me there. I have the same beef with Archimago. If he wants to be authoritative, he needs to use his real name. Objectivity needs to apply to him as well. People examine my background all the time in judging what I write. Why should Archimago be exempt? It was a mistake for Chris to use him as an expert witness while he was using an alias. Still is. So don't bring this argument to me. I am firmly on the side of full disclosure when it comes to people in the public/industry. Have always used my real name and will continue to do so.

 

[mtristand]

I don't understand this criticism. It is not their problem that people can't follow the discussion and generalize to lay understanding. Let me give an example.

The front end of a lossy codec like MP3 is indeed, lossy. It will transform the music to frequency domain, apply psychoacoustics, and generate a new bit stream. This bit stream still has redundancy in it so it is then fed to the back-end that is a totally lossless encoder. It uses a set of code books to assign smaller number of bits to common values, and longer ones to less common ones. So we get further reduction in file size. This can for example deal with silence in a file. And it is completely reversible when the decoder gets the bits.

You can't take what I just explained to say, "oh Amir just said MP3 is lossless." It would be the fault of the non-technical reader to do that, not mine trying to explain how the algorithm works.

And really, in this context, in this forum where we know what the terms mean, I don't understand why this is constantly brought up. If you now know what they mean, we are done with the topic. Constantly bring this up over and over accomplishes what?

Normally I might agree with you, but there are several more examples of questionable claims using the word "lossless" that I quoted, not just this one. If this video were the only example of the word being invoked, I might give them the benefit of the doubt. But it's not, so I don't.

To address this specific point: Yes this is exactly how I would expect MQA to respond as well - that technically there are aspects involved in the process that, in themselves, happen to be "lossless." But this is very different from claiming that overall, you have a lossless process where you can restore the original file in full.

No one would feel the need to question their intent if it weren't for all the other misleading claims that, when taken holistically, paint a picture of acting in bad faith. They do interject the word "lossless" all over the place in weirdly-worded ways to specifically take advantage of limited layperson understanding and give off a different impression to them.

It's always easy to fall back on the plausible deniability of saying "Well you just lack the technical expertise to read it carefully enough and that's on you." After all that would be the whole point of such wording - to be able to say precisely this, as a retort to any critic who "misreads." It's blaming the consumer for not being savvy enough to parse specifically-worded language that could have be stated much more clearly.

As a consumer I shouldn't have to be an expert who can pick apart tricky language. It is easy enough for the company to say "MQA isn't lossless, but what we offer is a transformation of what you do hear so that it sounds even better." Instead we get very carefully-worded claims involving "lossless" everywhere (including a very direct "Is MQA Lossless? Yes") to confuse people into thinking that they are somehow offering a lossless encoding solution overall.

 

[blackmetalboon]

You have no case with me there. I have the same beef with Archimago. If he wants to be authoritative, he needs to use his real name. Objectivity needs to apply to him as well. People examine my background all the time in judging what I write. Why should Archimago be exempt? It was a mistake for Chris to use him as an expert witness while he was using an alias. Still is. So don't bring this argument to me. I am firmly on the side of full disclosure when it comes to people in the public/industry. Have always used my real name and will continue to do so.

I have defended you and ASR on forums because I believe, in the long term, it will help dispel a lot of the nonsense in this hobby which will benefit us and future audiophiles in the long term.

 

[levimax]

I donated $20 to the MQA test fund

 

[dc655321]

I have the same beef with Archimago. If he wants to be authoritative, he needs to use his real name.

Were there issues with Archimago's analyses in this case? If so, those ought to be addressed.
Otherwise, his name, background, and profession are irrelevant to the technical discussion and being "authoritative".

BTW -- his identity is not difficult to discern...

 

[Raindog123]

Are you new to this discussion?

@amirm, I am not [entirely new here…] Yet am not aware of ‘anti-MQA PR talking points’. Where are they?

@All Here is a [rather sad] story from my life… I am ‘as technical as they get’. So, one day my former boss approached me with a request to be an expert witness in a large contract-award protest case, on technical merits. I agreed to look at the details and those did not pass my ‘moral compass’, so I refused... But it was a really important and big case for my boss, and did I mention I was the best guy for the job? So, he offered me a $100k bonus on the spot. I laughed and walked away…. But he came next day, with $200k… Long story short, half year later his company won the protest hands down. And I and my wife got two brand-new Range Rover HSE’s…. and I lost some respect by a few of my peers, and even some friends. But the Range Rovers were top-notch!

I do not really care about MQA. I’ve formed my opinion about it - technically, a ‘neat idea of the past’. Without neither a today’s purpose nor proof of proper implementation. Business-wise, just another startup trying to establish themselves on the market, while not having a ‘killer product’ so instead employing more-than-average aggressive, even cut-throat marketing... Well, just like the rest of the corporate world. So, if folks are willing to pay their ‘$5 a bottle’ while a nice cool $1 bottled water is offered right next door, I can only be sorry for them. But speaking of water, ‘you can only bring a horse to water, but you can’t make it drink’….

And Amir definitely deserves to cash-out his hard work of putting ASR together and building his reputation through it. But is it worth it - if it comes with a need to threaten prominent ASR members - the very fabric of the ASR - with ‘reply bans’ and with pushing them away?!

And BTW I too would gladly offer $250 toward proper MQA tests.

 

[mtristand]

You have no case with me there. I have the same beef with Archimago. If he wants to be authoritative, he needs to use his real name. Objectivity needs to apply to him as well. People examine my background all the time in judging what I write. Why should Archimago be exempt? It was a mistake for Chris to use him as an expert witness while he was using an alias. Still is. So don't bring this argument to me. I am firmly on the side of full disclosure when it comes to people in the public/industry. Have always used my real name and will continue to do so.

Some people will withhold their real name specifically because they don't want to introduce the possibility of a character/credential attack to distract from the actual argument in question, or because they don't want to get harassed by people with nefarious intent, etc.

If the argument itself is easy to dispute, then why not dispute the argument on its merits? It shouldn't matter where it comes from unless it's a suspicious argument not grounded in reality that can't be reproduced by others.

 

[jensgk]

But is it worth it - if this comes with a need to threaten prominent ASR members - the very fabric of the ASR - with ‘reply bans’ and pushing them away?!

It was not just a threat, he was actually reply banned.

And BTW I too would gladly offer $250 toward proper MQA tests.

I would not any longer. If MQA wants to regain some sort of decency and respect, they would have to pay for the tests and proofs.

 

[amirm]

Otherwise, his name, background, and profession are irrelevant to the technical discussion and being "authoritative".

So you say yet in every other post I have to field and insult or personal remark about me, Bob Stuart, MQA, etc. OP even went after my company with conspiracy theories for heaven's sake. Didn't see you coming with that defense for us. So have your standards and I will have mine. If you are going to write critical reviews of products, you need to do it under your real name. That is the precedence and is the right thing to do.

 

[amirm]

Some people will withhold their real name specifically because they don't want to introduce the possibility of a character/credential attack to distract from the actual argument in question, or because they don't want to get harassed by people with nefarious intent, etc.

Then don't jump in the middle of a hot potato called MQA. There are a ton of other topics you could write about. But when you anoint yourself a gladiator and jump in the ring, you best say who you are so we know how much you know, if you have connections to the relevant parties, etc.

If the argument itself is easy to dispute, then why not dispute the argument on its merits? It shouldn't matter where it comes from unless it's a suspicious argument not grounded in reality that can't be reproduced by others.

Which is what I have been doing yet every other post is personal again. About me, Bob Stuart being a snake oil salesman, etc., etc. So clearly you all don't believe that and think we are fair game personally.

 

[mtristand]

It appears that for most people the concepts of data space and musically relevant information are impossible to understand. And if you get stuck here, there is no path forward - what MQA does will forever seem suspicious and LOSSY.

I mean, this is the same discussion that happens with any lossy compression: How much (and what kind) of the relevant information can be retained while compressing the filesize?

You may argue that it's "perceptually indistinguishable from lossless" in some contexts, which is a separate argument - but either way, if the compression can't be reversed to the original, it isn't lossless.

 

[John Atkinson]

If you are going to write critical reviews of products, you need to do it under your real name. That is the precedence and is the right thing to do.

At the risk of inflaming the argument, if you are going to publish opinions and test results that may result in financial consequences for those whom you write about, you must do so under your actual name. With one exception - an exception that was agreed to before I joined the magazine - every contributor to Stereophile signs his work with his real name.

John Atkinson
Technical Editor, Stereophile

 

[mtristand]

Then don't jump in the middle of a hot potato called MQA. There are a ton of other topics you could write about. But when you anoint yourself a gladiator and jump in the ring, you best say who you are so we know how much you know, if you have connections to the relevant parties, etc.

At the risk of inflaming the argument, if you are going to publish opinions and test results that may result in financial consequences for those whom you write about, you must do so under your actual name. With one exception - an exception that was agreed to before I joined the magazine - every contributor to Stereophile signs his work with his real name.

He went into great detail here about this subject: http://archimago.blogspot.com/2018/10/musings-on-rmaf-2018-mqa-talk.html

Plenty of people do know who he is, and if he were being dishonest about his background or connections, it would have come out by now.

Either way, "how much you know" shouldn't be relevant in this case - do you disagree with the analysis he performed? Do you think he made any errors in methodology? Do you disagree with his results? Do you get something different when you try to answer the same question? If so, what did you do differently?

An argument relying on reproducible test results stands on its own merits and is a direct critique without introducing argument from authority.

 

[lucretius]

It appears that for most people the concepts of data space and musically relevant information are impossible to understand. And if you get stuck here, there is no path forward - what MQA does will forever seem suspicious and LOSSY.

Huh? There's no question about it. It is "lossy".

 

[KeenObserver]

You have no case with me there. I have the same beef with Archimago. If he wants to be authoritative, he needs to use his real name. Objectivity needs to apply to him as well. People examine my background all the time in judging what I write. Why should Archimago be exempt? It was a mistake for Chris to use him as an expert witness while he was using an alias. Still is. So don't bring this argument to me. I am firmly on the side of full disclosure when it comes to people in the public/industry. Have always used my real name and will continue to do so.

So true!
All the testing, graphs, and charts that you post here would be totally garbage if we did not know your true name.