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

[filter_listener]

The issue is, the tests that GoldenOne did show clearly that MQA fails abysmally at all of (A), (B) and (C).

Would you like to be the one to address the points in misewall's post since everyone else seems to be ignoring it?

 

[dmac6419]

You want white papers? I could point you to some "proving" that they have made super-luminal cables, or a coating for a resistor that reverses entropy.

I can point myself,but thanks for help,have a happy Saturday

 

[Jimbob54]

The issue is, the tests that GoldenOne did show clearly that MQA fails abysmally at all of (A), (B) and (C).

It fails miserably as a dirt racer too, according to @mieswall . Or something like that. https://www.audiosciencereview.com/...-music-on-tidal-to-test-mqa.22549/post-749270

Begs the question, what is it good for?

Well.... https://www.audiosciencereview.com/...-music-on-tidal-to-test-mqa.22549/post-748598

 

[voodooless]

A) Improve end-to-end time domain reproduction by utilizing b-spline convolution & resampling + complementary filtering(1)(2)(3)

B-spline interplolation does not adhere to the sampling theorem, and the results shown here and also by various others have shown that very clearly. I still have no idea how they intend to compensate for the various apparent time domain issues that were introduced while mastering?

B) Reduce the file-size required to store high-resolution audio(4)(5)

Who cares? Tidal, because they use less bandwidth? This is just a very poor argument.

C) Provide provenance information for the file to verify that the audio has not been altered in transit (7)(8)

Which has been proven bullshit.

If you don't find any of these features compelling... don't use it!

Thanks, I won’t!

 

[mansr]

Would you like to be the one to address the points in misewall's post since everyone else seems to be ignoring it?

Ignoring it seems like the most polite thing to do.

 

[sandymc]

Would you like to be the one to address the points in misewall's post since everyone else seems to be ignoring it?

If you want. misewall's post can essentially be summarized by "MUSIC oriented algorithm it is". That's fine. There's a place for perceptual encoding of music. About 99% of the population are happy with MP3s. Anyone that feels that that MQA sounds better that lossless encoding, or the same, is more than welcome to use it. The issue is MQA has been claimed to be lossless, authenticated, etc, etc. And it is nether lossless or authenticated. The problem is, people are paying good money for something that doesn't do what is claimed.

 

[Tks]

Would you like to be the one to address the points in misewall's post since everyone else seems to be ignoring it?

What's there to address seeing as how all he did was specifying precisely on the ineptitudes of the encoder itself has as far MQA explains, and no explanation of how to verify the benefits it purports are present within "real music" or generally within any of the product's beneficial aspects to consumers?

I could ask you the same thing though. Can you explain the other things like how you could confirm any of their patent content is actually what is being delivered on seeing as how the codec's version iterations have changed, doing who knows what? Can you explain why the blue light is on, after chopping up an MQA file for nearly 30% of it's content through truncation? Can you explain why OP's music was offered in 88.2 when the "master" was 44.1? Can you address how it can provide evidence of provanance of what ADC was used and taken into account for OP's music being MQA encoded and the supposed DSP applied with this in mind to create whatever vacuous claim of superiority over a lossless file? Can you address why they won't open up to NDA'd third party researchers for audits at the very least? Can you explain why ALL test tones are failures, yet "real music" doesn't and I mean very specifically without human intervention or the post-hoc little bullshit claim they make how the encoder is failing when all it is, is a simple test-tone detector present to avoid embarrassments like this, surly one test tone method can survive not devastating the encoder? Lastly, if the encoding failed, why was the music even published (EVEN AFTER there was indication OP was trying to pass off test-tones the first time), how on EARTH can that conform with the "original" provenance claims MQA posits it can maintain? If there was a failure as claimed by the encoder, why would failures be published without any communication to the artist?

I could keep asking about the hilariously bad damage control attempt, and why they're cucking away into the shadows by having OP's music removed (demonstrating the pull an encoder developer can have on a publisher, and in-turn Tidal as a platform, which is insane when you think about it). But I'll settle for answers to the prior stuff.

 

[dmac6419]

If you want. misewall's post can essentially be summarized by "MUSIC oriented algorithm it is". That's fine. There's a place for perceptual encoding of music. About 99% of the population are happy with MP3s. Anyone that feels that that MQA sounds better that lossless encoding, or the same, is more than welcome to use it. The issue is MQA has been claimed to be lossless, authenticated, etc, etc. And it is nether lossless or authenticated. The problem is, people are paying good money for something that doesn't do what is claimed.

I'm paying the same money I paid before and after,I don't see your point,I know for a fact that the music label will charge you more for 16/44 24/44-48 24/96 24/192 about a dollar or two for each resolution you go up (if you want to own it),music ain't free son,people have to eat.$20 or less is nothing for streaming access to millions of music files for whatever streaming services you decide to use.

 

[filter_listener]

Ignoring it seems like the most polite thing to do.

@amirm Is this the kind of audio forum you want to be running? I'm trying to have a substantive conversation, this is what I'm getting in return.

 

[sandymc]

I'm paying the same money I paid before and after,I don't see your point

Actually you aren't. A MQA capable dac is more expensive than the non-MQA equivalent due to licensing costs. E.g., Topping D90 vs D90 MQA

 

[threni]

@amirm Is this the kind of audio forum you want to be running? I'm trying to have a substantive conversation, this is what I'm getting in return.

Well, this is the sort of forum I want him to be running. You can report posts you believe breach the TOS here.

You want him to compel people to answer any question anyone poses? Why? Why would he want to do that?

 

[Jimbob54]

@amirm Is this the kind of audio forum you want to be running? I'm trying to have a substantive conversation, this is what I'm getting in return.

Well, at least he was civil...

 

[GoldenOne]

Would you like to be the one to address the points in misewall's post since everyone else seems to be ignoring it?

I've already addressed the concerns, and have said that I would be quite happy to work with MQA to conduct testing that would be agreeable to both them and myself.
But given as they refuse to allow any third party testing and the criteria by which they deem something to be within the bounds of the encoder's capabilities are vague to say the least (who the hell decides what is/isn't 'natural sound'??) there is nothing we can do.

I'd be happy to do further testing if they work with me. I tried to test with low level, and more 'natural'/less demanding signals, but they nuked the track before it was published, so I wasn't able to show the results.
If that track provided better results than the ones i've shown I would have been more than happy to speak positively about that and agree that within certain limitations maybe MQA delivers on some promises.

But they prevented me from doing that, have provided no evidence for it themselves, and so all we can do is go based off existing testing.
If the MQA encoder can't handle a sine properly, I have no faith in it for music.

I do believe that there are many situations where sine/steady state signal performance of a product or technology may be better than the musical/'random' signal performance. But the other way around? I doubt that.....
Happy to be proven wrong. That's the scientific method after all.

But i'm not happy to blindly accept marketing claims when all evidence so far contradicts them. Even the musical content in my tracks was butchered.

 

[dmac6419]

Actually you aren't. A MQA capable dac is more expensive than the non-MQA equivalent due to licensing costs. E.g., Topping D90 vs D90 MQA

Not mine, look at the mola mola mine measure better and cost thousands less and has MQA,d90 and d90 mqa cost thousands less too.

 

[mieswall]

I never claimed to understand the inner workings of MQA. In fact that point was expressed in the video. Half the frustration is that despite the "Explanations" and paper written by MQA, they go to great lengths to restrict anyone from testing the claims.

One also does not need to fully understand something to test simple claims about it. You don't have to fully understand the entire processes within a delta sigma DAC to demonstrate that it has good/bad objective performance.

MQA themselves were also given the opportunity to respond, which in my opinion they did quite poorly. And they have not responded to my (or others) requests to work together to conduct testing that both sides agree upon.

I did "do my homework" and read through all available material on MQA I could possibly find. And whilst much of it provides explanations or claims, it is not proof.

Anyone can patent anything regardless of if it is effective, or of any use https://patents.google.com/patent/US20060094518A1/en
And anyone can write a paper, it again is not proof. It is then up to others to either support it or come up with evidence that contradicts it.

I tested their claims, they showed to be false. Now it is up to them to provide or allow others to collect evidence that shows that for normal music it works as they claim.
Until that happens, there is no reason to trust them.

I'm surely not even close as tech savvy as you, but it is hard to deal the fact that people like me or everyone truly interested in this matter can understand the basics of MQA just reading the same public information available, while you don't.

One example: MQA states in their graphs and techie explanations that they capture a "triangle" of significant data (the specific shape of it as previously analyzed by the algorithm previous to coding -that's why even you mentioned the "electronic music" option for coding-, and btw based in a most fundamental profile of the harmonic content that every musical instrument has, that people like Bach discovered centuries ago), and whose limits are in its upper side the maximum amplitudes at each frequency (almost always diminishing the further you go in frequency.... or you will tell us that you have never seen a FFT of an instruments' note?), and in the lower limit the noise floor of the combined effects of the multiple steps of digitalization equipment or later, the reproduction chain.

Then, what's inside that triangle of significant data is captured, *in a lossless way, there that matters*, and what outside is not (and that's what can be named "lossy" if you are taliban enough, because it purposely forgets the noise and the alleged "data" above music). Then what's not inside is gained space is better used for receiving data coming from upper origami's folds (where the same process occurs, but in a even more bold fashion, as the true data is even less, and so the significant space of true data is much more smaller). That obviously means, as @filter_listener said, a variable bitrate in frecuency. Why use 24 bits depth in, say, 15 Khz, while the significant musical data only spans for 5 to 6 bits there, while the rest is silence or noise? If I use the remaining 18 bits for other more useful purposes than recording that noise floor, is it that a sin?

And then, knowing that (sorry, I simply can't believe you don't), you feed your test tones full of noise and square waves, that obviously place a lot of high amplitude content where you already know that MQA will not properly code it, because it will fall outside the space where MQA is designed to code. Exactly as if you were dealing with a standard compression algorithm that could deal with images, music, or whatever. What is the purpose of that? What useful conclusion can your readers have if you mislead their opinions pointing at things like: look! what a disaster! those rookies of MQA (no other than the same guys that invented the first useful lossless compression schemes!) can't properly rebuild a square wave, because they are rendering 5 o 6 bits at that given frequency, instead of the 15 bits needed to properly reconstruct that square wave with those higher harmonics!

If instead you don't know, which seems to be your current argument ("I never claimed to understand the inner working of MQA") those basic characteristics of the MQA process, how you even dare to qualify them, let alone measure it like if it were a common compression algorithm?
(sorry for my "lossy" English, I hope you understand the idea....)

 

[GoldenOne]

I'm surely not even close as tech savvy as you, but it is hard to deal the fact that I and everyone truly interested in this matter can understand the basics of MQA, reading the same public information available, while you don't.

One example: MQA states in their graphs and techie explanations that they capture a "triangle" of significant data (the specific shape of it as previously analyzed by the algorithm previous to coding -that's why even you mentioned the "electronic music" option for coding-, and btw based in a the most fundamental profile of the harmonic content that every musical instrument has, that people like Bach discovered centuries ago), and whose limits are in its upper side the maximum leves at each frequency (almost always diminishing the further you go in frequency.... or you will tell us that you have never seen a FFT of an instruments' note?), and in the lower limit the noise floor of the combined effects of the multiple steps of digitalization equipment or later, the reproduction chain.

Then, what's inside that triangle of significant data is captured, *in a lossless way, there that matters*, and what outside is not (and that's what can be named "lossy" if you are taliban enough, because it purposely forgets the noise and the alleged "data" above music). Then what's not inside is gained space that is better used for receiving data coming form upper origami's folds (where the same process occurs, but in a even more bold fashion, as the true data is even less, and so the significant space of true data is much more smaller). That obviously means, as @filter_listener said, a variable bitrate in frecuency. Why use 24 bits depth in, say, 15 Khz, while the significant musical data only spans for 5 to 6 bits there, while the rest is silence or noise? If I use the remaining 18 bits for other more useful purposes than recording that noise floor, is it that a sin?

And then, knowing that (sorry, I simply can't believe you don't), you feed your test tones full of noise and square waves, that obviously place a lot of high amplitude content where you already know that MQA will not properly code it, because it will fall outside the space where MQA is designed to code. Exactly as if you were dealing with a standard compression algorithm that could deal with images, music, or whatever. What is the purpose of that? What useful conclusion can your readers have if you mislead their opinions point at thing like: look! what a disaster! those rookies of MQA (no other than the same guys that invented the first useful lossless compression schemes!) can't properly rebuild a square wave, because they are rendering 5 o 6 bits at that given frequency, instead of the 15 bits needed to properly reconstruct that square there for those higher harmonics!

If instead you don't know, which seems to be your current argument ("I never claimed to understand the inner working of MQA") those basic characteristics of the MQA process, how you even dare to qualify them, let alone measure it like if it were a common compression algorithm?
(sorry for my "lossy" English, I hope you understand the idea....)

As said, I published a track that conformed to what MQA (from what they've said, including this triangle) deems 'natural'.
They nuked it.

I said in the video, I'm happy to update the posts etc if they send me the MQA version of that file and it is indeed much better than the others.
They have not done so.

If you disagree with my testing, then I would strongly encourage you to conduct your own.
The world would probably advance much faster if people spent as much time developing their own ideas as they did poking holes in things put forward by others.

 

[sq225917]

Does it strike anyone else as odd that such a contentious topic brings a glut of new posters?

I've never seen as many odd socks on the floor in a thread....

 

[filter_listener]

Can you explain why the blue light is on, after chopping up an MQA file for nearly 30% of it's content through truncation?

As already mentioned, 16-bit truncation is explicitly allowed for by the MQA decoder: https://bobtalks.co.uk/blog/provenance/provenance-and-containers/

Can you explain why OP's music was offered in 88.2 when the "master" was 44.1?

What? The MQA Core Decoder works at 2x rate always, this isn't a secret.

Can you address how it can provide evidence of provenance of what ADC was used and taken into account for OP's music being MQA encoded and the supposed DSP applied with this in mind to create whatever vacuous claim of superiority over a lossless file?

No, not with 100% accuracy. I wish that it werent this way and have already expressed my desire for more information from MQA about how this works.

MQA says that they've profiled all of the common ADCs and filters and can identify with some level of accuracy what effect they're having on the audio so that it can be corrected and compensated for. My guess is it's related to this patent, it seems plausible that they're identifying patterns of pre-ringing in the audio signal, applying correction, and then embedding what compensation filter to engage for downstream decoders/renderers. This isn't rocket science, but I'm sure I'm missing the nuances of it.

Can you explain why ALL test tones are failures, yet "real music" doesn't and I mean very specifically without human intervention or the post-hoc little bullshit claim they make how the encoder is failing when all it is, is a simple test-tone detector present to avoid embarrassments like this, surly one test tone method can survive not devastating the encoder? Lastly, if the encoding failed, why was the music even published (EVEN AFTER there was indication OP was trying to pass off test-tones the first time), how on EARTH can that conform with the "original" provenance claims MQA posits it can maintain? If there was a failure as claimed by the encoder, why would failures be published without any communication to the artist?

It's not a "failure" of the encoder, it's a protection they've implemented to prevent people from feigning outrage by getting unexpected results and interpreting them incorrectly. Exactly the scenario that's happened here.

Having different encoding parameters for different types of content is commonplace in the video world... I'm unsure why this is so controversial in audio. Search for "Anime video encoding" or "Anime video upscaling", using generic video processing on Anime or vice-versa often results in sub-par and/or hilariously broken results. Here's an example: https://netflixtechblog.com/per-title-encode-optimization-7e99442b62a2

 

[mansr]

Does it strike anyone else as odd that such a contentious topic brings a glut of new posters?

I've never seen as many odd socks on the floor in a thread....

Those are not just any old socks, they are sock puppets.

 

[PierreV]

And then, knowing that (sorry, I simply can't believe you don't), you feed your test tones full of noise and square waves, that obviously place a lot of high amplitude content where you already know that MQA will not properly code it, because it will fall outside the space where MQA is designed to code.

If anything, this kills the MQA lossless claim. I know that your answer will be that MQA has admitted it is not lossless. However, you can't dispute that it is advertised as lossless by many, if not all, MQA partners (if not, at times, as better than lossless, closer to the intended experience).

That marketing is obviously misleading (to be kind), as demonstrated by the above.

I can't wait for MQA to reign in those misleading advertisers who are currently tarnishing its image.