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

[DawgSlaya]

It seems like a simple matter to me:
I don't have access to an analog eq, thus I value my digital eq more than whatever this deblur process is that MQA is employing.
After all there is a very audible difference with and without headphone corrections.

 

[restorer-john]

Full Ack. Wind chimes is really hard stuff for anything.

Chimes (metal bar) find their way into a lot of smooth jazz recordings, often as highlights in introductions and concluding parts of tracks. The good ones stand out, whereas the bad recordings sound like a splattered mess with no detail.

Way back I first heard this splatter on a Technics portable CD player, and tracked it down to the 40 second anti-shock RAM. They were using a lossy encoder/decoder to extend the antishock time without putting in a more expensive and larger RAM chip! When switching off anti-shock the distortion was gone, put it back on, and chime splatter. For other music, you couldn't tell.

That tiny excerpt became a test track for me.

 

[JohnYang1997]

they could easily define it as perceptually lossless, which is the goal of any audio codec in the world that isn't called FLAC.

Cool. Assuming this as an valid explanation. How do I get my head around better than lossless. That doesn't seem right.
From my understanding, it could basically mean CD quality as most lossless files implied to be.

 

[abdo123]

Cool. Assuming this as an valid explanation. How do I get my head around better than lossless. That doesn't seem right.
From my understanding, it could basically mean CD quality as most lossless files implied to be.

it would be better than lossless if you would take their other claims for face value .

 

[danadam]

You have to keep the hypothesis of the video in mind: MQA is lossless.

From https://bobtalks.co.uk/blog/mqaplayback/origami-and-the-last-mile/

MQA uses a process we call ‘Music Origami’ that focusses on maintaining the information in the orange triangle (plus a substantial safety margin)

Therefore MQA is not lossless.
In addition, IMO it is safe to assume, that the "safety margin" does not extend beyond the gray rectangle (i.e. CD spec), otherwise they would surely mentioned it, and therefore MQA is not lossless even for the 16/44 part of the hires file.

QED and no youtube video was necessary

 

[voodooless]

Not at all. There is for example "perceptually lossless."

Really, that is the best you can come up with? The FAQ question is and was not: "Is MQA perceptually lossless?".

This is when you tell a lossy encoder to produce the highest quality it knows, and have the bit rate vary. This is as special VBR mode that is available for some lossy codecs. We had this at Microsoft and in vast amount of content it was transparent and about half the size of mathematical lossless.

So you'd be fine if Spotify also claims in their FAQ that they are lossless? Obviously, they can fluff up the answers with some meaningless drivel.

MQA proposes another form of lossless that says it music follows statistical distribution, they are able to fully preserve its actual informational content including what is in ultrasonic. If they are able to achieve this fully, what is your beef with it? That it doesn't spit out a bunch of bits for a rectangular channel is not used in music?

I don't have any beef with that. I have beef with the unclear messaging and several claims that cannot be backed up or verified. I don't mind a lossy codec.. I use Spotify.. so sue me

When we want to say the same bits come out that go in, we clarify with the term "mathematically lossless."

In 99% of places we know that when it says "lossless" we mean "mathematically lossless". If MQA says lossless, we suddenly need to use another definition.. That is quite baffling to me.

Remember, there is no lossless codec that works for all content.

Sorry, but that is just a deflection tactic. It's just the degree of compression that is different, so it's only an efficiency question. They are all lossless, regardless of what the input is. There is no lossless codec that tells me I cannot compress a specific file because it's the wrong content. if you want to put a word document to the flac encoder/decoder, that would work just fine (given that you would add the needed headers to make it think it's a WAV file). No, it will not compress very well, but it will work, and it will be perfectly lossless.

 

[restorer-john]

if you want to put a word document to the flac encoder/decoder, that would work just fine (given that you would add the needed headers to make it think it's a WAV file). No, it will not compress very well, but it will work, and it will be perfectly lossless.

Yeah, but it will make the MQA encoder explode.

 

[tuga]

This sounds a lot like "mp3 reloaded"...

Is MQA (losslessly) higher-res than CD-quality/Redbook? (Redbook is lossy of higher-res content)

We don't know whether or not it is, and cannot find out because MQA is a secret.
If it isn't we don't need it.
An 19-bit/88.2kHz FLAC (level 8 compression) takes up less space and it doesn't rely on some perceptually-based algorithm to chop-off what it believes I won't hear or doesn't matter musically, nor does it apply some leaky filter.
And the music file player/software I bought/paid-for can play it and add EQ.

 

[Deleted member 13028]

MQA is lossy. I know more about this than anyone here as I have spent a whole decade losing stuff.

 

[stopbeingalentil]

“To help young artists and small labels get their music encoded in MQA and on to TIDAL, we recently enabled the service you used. However, that service is limited in flexibility and places obligations on the user. First, the encoder is fully automatic, which means it will use analysis to set parameters for each song as a whole; second it is intended strictly for music. This encoder is not configured to deal with content where, for example, the statistics change mid-song, or where the audio does not resemble natural sound. The onus is on the submitter to check the content when it arrives in TIDAL and confirm the sound. In this way, we can all be sure that the provider is happy with the Master and that, because of the light, everyone with an MQA decoder is getting the intended sound.”

This is not marketing.

How is this not marketing? Like, who decides what's music or "natural sound"?

You were clearly told why the standard encoder (or any of their encoders) are not designed or optimized for encoding pathologically test signals. And that the encoder that was used for your content was not the optimized one they use.

This has been the same point I and others have been making.

Given this information, OP should have clearly stipulated that a different encoder was used than one for major labels and that he was warned the encoder would not work for his content.

This would have been speculation - nobody knows which encoding process is behind which release or if different labels get different processes or what. More to the point: The fact that whole back-catalogs get made available now in MQA indicates that there is in fact an automated, mass conversion process behind many MQA releases, which in turn is only feasible with a generalized encoder. That may or may not be the same that was used with Golden's tracks but at any rate it's almost certainly not tuned for the majority of releases it has been applied to so far.

Also he was not warned about it. Had he not contacted MQA about this, he'd not have gotten any technical feedback on the tracks that weren't outright rejected right away.

 

[stopbeingalentil]

At this point, it has to be a meeting of both parties/camps' minds, moderated by an independent trusted figure(s). As without this attempt - to get things (tests) right - no matter what you, Bob (or I) will say will sound and smell like smear campaign and tracks covering.

If MQA had anything to gain by a more open approach, they would have done that by now. Consequently it's not unreasonable to assume that any findings coming out of that would be to the detriment of MQA. Now, why would Amir put his relationships with different industry individuals on the line by facilitating that when MQA already got to the point of having virtually all relevant DAC manufacturers by the balls, the backing of most major labels because of promises of better audio fingerprinting and DRM schemes (my personal interpretation), and already having reached critical mass for adoption and customer acceptance - heck, even customer brainwashing - as it's customers who push manufacturers to include MQA in their offerings even if the manufacturers hate it?

 

[stopbeingalentil]

And these are the guidelines for AES Journal: https://www.aes.org/journal/authors/guidelines/

"All submissions will go through a peer review process to check their suitability for JAES. "

That paper has whopping 150 references in it! That doesn't make it necessarily right but it is the level of effort required to get something published in the Journal.

This is the danger in mounting such strong challenges without having the prerequisite knowledge of the field.

Are you serious?

Unless they name the people who peer-reviewed that paper, this claim is laughable IMHO. It has been shown time and again, in all different kinds of scientific fields, how journal "peer reviews" can often be but a bad joke, with even reputable journals regularly including "peer reviewed" papers that had been entirely computer generated nonesense (ie. a glorified, scientific sounding lorem ipsum text).

Trying to find out something about the credentials of the co-author also comes out empty aside from the few lines on him at the end of the paper. For a company who regularly pushes the "credentials rethoric", this is pretty thin.

 

[Lunafag]

Really sad to see how money can make people defend the existence of a ****** lossy codec with DRM and royalties. We already have fantastic royalty-free codecs, both lossy and lossless.

 

[stopbeingalentil]

It took a long time for the technical side of MQA to be understood broadly by the community. I mean by a lot of people, not by the few who are in a position to understand through technical training.

This is wrong. Since the MQA process is closed and proprietary, the only people able to understand it in full are MQA themselves. Everyone can only chose to believe what they claim.

All that was based on research and good understanding of musical content and listener conditions.

So they claim without providing any proof of it, without stating their testing methodolog. At the same time they actively work against independent efforts to test their claims.

At any rate what stands, imho, is that MQA is to no benefit for the consumer whatsoever.

 

[stopbeingalentil]

So, if I understand correctly, MQA could be lossless, or with a very small loss (whatever this may actually mean) but only if the signal resembles an average musical signal.
The encoding in MQA format can and should be "tuned" for the specific music being encoded for the best artistic result.
Since it may well seem, from the marketing material, that MQA is a lossless compression, the Golden One wanted to test this point.
The tracks used, though, were non "musical" in content and the result was that the output was, indeed, different from the source.
This turned out to be against the terms of service so the content was removed.
Since the Golden One made a public video, a public reply was issued.

If I am summarizing correctly, we learned something interesting about MQA. The test that is missing, correct me if I am wrong, is to do the same with a conventional musical track and see what happens and where are the differences. Maybe we can find some musicians who may want to do the test. That should not be against the terms of service, I suppose.

Firstly: MQA could never be lossless when they claim to "improve" anything. Being lossless means being identical to the original output. If you change anything about that, then you're not lossless any more.

As to MQA's influence on real music:

Here's a spectrum difference of "Poppy"'s "Concrete", which is available as MQA on Tidal. The comparison is between the 44.1/16 file from Qobuz and the 44.1/16 file from Tidal (ie. "Hifi"/non-mqa)... This both shows that you still get the MQA encoded file even if you select "CD-Quality" in Tidal and how the MQA encoding fucks with the sound even in the audible band. Whether that's generally audible is a different question but the difference is significant nonetheless.

 

[stopbeingalentil]

Can Ford sell you a car with experimental firmware someone in the assembly line put in there to see what happens to the car when you drive it?

I'm pretty confident that Tesla does this all the time. BTW so does Microsoft with Windows, Google with Chrome, Mozilla with Firefox, and most SaaS providers with their services.

 

[Vladimir Filevski]

Unless they name the people who peer-reviewed that paper, this claim is laughable IMHO.

There is no science journal in the World which reveals the names of the reviewers - that is the nature of the peer-reviewing process accepted by the whole science community. Please accept this fact. Or if you can not accept it, please do not comment on this.
Of course, there are a few examples where reviewing process was not good enough, but because the science is self-correcting, all of those accidents were quickly revealed.
Edit: The Journal of AES has never had such accidents, ever!

 

[jensgk]

And you would be mistaken in that if you had read other things he says about us in other fora. Here is an example of his posts:

View attachment 132333

Obvious ties to SMSL/Topping? They have it in everything??? You think anyone saying that about me is an upstanding citizen?

What he quoted after that had nothing to do with "pushing DRM schemes." It was an email I sent to Bill Gates letting him know that we were trying to get our hardware ecosystem of players to be as good as Apple's which had seamless Fairplay DRM integration.

Goldeneye proceeds to more sensational conspiracy theories.

Adam took an initiative to call him and offer to have a new thread where I would NOT post (as to keep tensions low) and he could start over. He went and disclosed that call with his own spin as a nefarious thing to do:

View attachment 132336

So yeh, we are to hire more moderators so that the food fights that he created can continue. And that I better say nothing or else. We go out of our way to allow dissent in this forum and to have it toiled this way is in very poor form.

Net, net, while he did some good work, I consider GodenEye to be a bad actor here. We are providing incredible visibility and traffic for his videos and yet he brings nothing but disdain towards us. Keep your guard up. Don't believe him because he sounds calm and collected in his videos.

@amirm, Within the many audio communities, you are a very public internet figure. Your expressions and motives can and will be discussed in many forums, especially given your strong defense of MQA, which is very puzzling to many of us.

 

[pjug]

I don't get why MQA is so determined to hold onto the lossless thing. It is lossy and they pretty much have conceded this with the scrubbing of their original "MQA Lossless". Seems to me they could just say "very best fidelity" or something like that.

Anyway, Tidal and some of the labels seem to have moved on from MQA to MQA-CD "masters". We are debating the right-hand quarter of the chart. Meanwhile the left 3/4 keeps growing.
(from the .csv found here: https://www.meridianunplugged.com/ubbthreads/ubbthreads.php?ubb=showflat&Number=268318)

 

[PO3c]

It is not definition of music. It is the spectrum of music in highest frequencies. Across millions of tracks, they have built a very good statistical profile of worst case amplitudes. It looks like this:

The brown graph is the *peak* level of all the content they have seen so far in encoding MQA content. OP's two test files are in olive and red colors. As you see, they wildly exceed the worst case spectrum of content they have seen across million+ tracks.

MQA when asked to encode high res files into standard res it not mathematically lossless. It cannot be. Question is, if music is encoded at its full information-level or not. They say it is and OP's tests don't nullify that.

Vast majority of you are in favor of 44.1 kHz sampling because you think there is nothing useful above 22.05 kHz. It is super odd now to get religion and say that what is above 22.05 has a) a ton of amplitude and b) must be preserved.

Am I wrong thinking this triangle very easily could be documented by Bob Stuart giving access to encoded files? Surely any sine wave could exist inside this triangle without artefacts?

Personally I'll be interested to learn how much damage are done to non unfolded files as this is how I'll be using MQA content forced on me.