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

[scott wurcer]

Son this is not a white paper

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.

 

[mansr]

You want white papers?

I have rolls of white paper by the toilet.

 

[Tks]

THE WHITE PAPER! BEHOLD! (Time stammped, just press play)

 

[filter_listener]

This is a false claim, that is bordering on hypocrisy, seeing as how GoldenOne's channel is one that of mostly subjectivist leaning. He recomends things like the M-Scaler which is basically laughed at here not due to the technical ability with it's one mill taps, but because it's largely seen as completely pointless, and at ridiculous cost if you pair it properly with the Dave.

I have no idea what GoldenOne/GoldenSound's background is, nor do I think it's relevant to the discussion at hand. I read the post about his MQA "test" and drew my conclusions based on the number of incorrect assumptions and conclusions.

Second, poor understanding of how MQA works? Everyone has poor understanding of how it works, because it's closed source nonsense, and whatever could be extrapolated only points to that nonsensical direction as presumed. No amount of Bob's discussion on how he approaches how mics and ADC's used in original recording of produced music will have baring on the product offering that MQA is.

Again, I suggest reading over the patents filled by Bob and Peter if you're genuinely curious about how MQA works. I agree that the material in the patents should be made in a more digestible format, but the information is there.

Lastly, you've simply asserted claims here about what he did, yet have not demonstrated any of supporting examples how you reached such conclusions about him. The sheer lunacy lost on them in not realizing this is more of an argument against their encoder is staggeringly hilarious. The man was creative enough to get passed considerable gating, and basically got MQA to admit their encoder is so garbage, it can't do test tones properly, but somehow we should still trust it works fine for actual music?

And you want to call someone like this some amateur? Why?

mieswall's post already summarized the significant deficiencies in the author's testing methodology and domain knowledge. MQA is not designed to encode test tones, it uses complicated and inter-depedent processes specifically optimized for encoding of the 99% of recorded music that humans listen to. It's not like MQA are opaque about this, as I linked in my post there are multiple sources explaining how MQA uses DSP in the encoding and decoding processes!

It's not my (or anyone's) job to explain this stuff to him... it's obvious to anyone who has "done the homework" that he didn't do it.

 

[mansr]

It's not my (or anyone's) job to explain this stuff to him... it's obvious to anyone who has "done the homework" that he didn't do it.

You're referring to Bob Stuart, right?

 

[jensgk]

@GoldenOne , I am very impressed with your analysis, and hope that Tidal and MQA will change course or perish.
I don't know if the following files are based on the same master, but it sort of is implied: Test Your System with Free Downloads from Blue Coast Music
Anyway the same piece of music freely available in DSD, WAV, FLAC and MQA. Maybe it could be used for further analysis?

 

[voodooless]

5- time smearing or blur is caused by three issues: a) the phase shift of harmonics produced mainly (but not only) by analog filters at capture

How do harmonics come to play here? What capture? There is a shitload of equipment involved in a studio. Probably multiple ADC’s, lots of DSP processing, probably at least a half dozen sample rate conversions. How can MQA compensate for all? If this were important, why would the studio not do the compensation themselves?

b) aliasing close to the Nyquist frecuency, corrected by MQA by adaptative filters;

Oh, You mean that massive band of HF noise that is still present even after decode? Why is phase response important in that area vs the same area being flooded with noise? Or are you eluding to the leaky filter that is being used for upsampling? Talk about aliasing…

c) ringing on impulse response, that MQA dramatically corrects using convolutional filters.

Oh yes, we’ve seen the impulse response of the MQA filters.. no thank you!

 

[GoldenOne]

I have no idea what GoldenOne/GoldenSound's background is, nor do I think it's relevant to the discussion at hand. I read the post about his MQA "test" and drew my conclusions based on the number of incorrect assumptions and conclusions.



Again, I suggest reading over the patents filled by Bob and Peter if you're genuinely curious about how MQA works. I agree that the material in the patents should be made in a more digestible format, but the information is there.



mieswall's post already summarized the significant deficiencies in the author's testing methodology and domain knowledge. MQA is not designed to encode test tones, it uses complicated and inter-depedent processes specifically optimized for encoding of the 99% of recorded music that humans listen to. It's not like MQA are opaque about this, as I linked in my post there are multiple sources explaining how MQA uses DSP in the encoding and decoding processes!

It's not my (or anyone's) job to explain this stuff to him... it's obvious to anyone who has "done the homework" that he didn't do it.

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.

 

[tmtomh]

Hey tmtomh, I was hoping you would see this new development seeing as how I recall in the recent past you had to keep explaining to a few people about how ridiculous MQA claims were at face value even conceptually without even having to do the stuff GoldenEars did.

What do you think about his expose'? Saves you from ever having to suffer frustrations explaining the basics about the blue light being on and why it's proves nothing.

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

@amirm Where you at bossman? I know in the past (I'm talking like years ago), I've seen your alias pop up on other places on the web challenging Bob to come out and demonstrate his claims after your long explanations on how MQA is basically making claims that cannot be true. Though just very recently, I recall there being an incident where you were a bit triggered or something when a few people were questioning your position on DRM-like natures and the work you did I MS back then to try and push for an authoritative position for an audio format that Microsoft would be in control of. And it somewhat sounded like you were for some reason coming to MQA's defense as a product that shouldn't be as detested as it is (which I recall asking about, but it seemed the discussion never properly concluded).

Now that this whole thing is somewhat settling. Whats' your position on MQA currently? You still like chasing MQA versions of music over lossless for the sake of finding different masters potentially? Also do you still hold to the notion of MQA being something we should support being an option on the market in it's current state? Really interested in hearing any conclusions you have on the matter since I find it odd you haven't spoken about this whole ordeal.

Thanks for the tag! I should say first of all that I was just one of a group of folks over at ComputerAudiophile/AudiophileStyle who were deeply involved in the arguments over MQA, and the commentary I have made here about it has been deeply informed by the work @mansr and other folks did over there in the gigantic MQA thread.

That said, yes, I was quite intrigued to see @GoldenOne 's in-depth analysis, which built on mansr's and others' work. I don't think that detailed analysis Golden One did is irrelevant or redundant at this point: it provides empirical evidence of what we've been saying; it has some surprising new results that don't alter the basic issues but do further our understanding of this black-box technology; and it also contains a great one-stop list of informational links.

I don't think, however, that all this data is going to necessarily convince folks who don't want to be convinced. After all, there are more fundamental, basic, easier-to-understand principles of digital sampling theory that explain why, for example, very high sample rates are pointless, but that hasn't stopped the "audiophile=highest res possible" equation from taking root in both hardcore audiophile circles and in more casual hi-fi circles (for example vinyl ripper communities and high-res/HD streaming consumers). It is a great resource to point to, though, and is tremendously helpful because we can now just show some stuff is instead of having to try to explain how it logically must be the case.

As for @dmac6419 's comment that @amirm won't be touching this, that's probably true. Let me first say I deeply respect Amir's willingness to let folks (including me!) get into it with him so strongly about MQA. While this is a science- and fact-oriented forum, it's still his forum, and he's allowed folks to basically go off on him, and directly at him, about this topic more than most owner-mods of other forums would. That said, if one reviews Amir's comments in the forum here about MQA, I think one will find they do not quite fit his usual pattern of clearly and unsparingly condemning bad technological implementations and snake-oil BS, which is what MQA absolutely, positively is. But not every individual, even Amir, has to be equally interested in or passionate about critiquing every aspect of our hobby. When someone else posts a test of a piece of equipment, no one asks why Amir isn't testing that equipment himself. Same with MQA as far as I'm concerned - if he doesn't want to get into that fight, other folks can. It's a community, and as long as Amir allows the criticism to be made here, there's no reason he has to be compelled to join in if he doesn't want to.

 

[filter_listener]

You're referring to Bob Stuart, right?

I genuinely think it's sad that you value trolling and toxicity over discourse and education. Scoring internet points obviously brings you joy and who am I to say whether that's a good thing or a bad thing... but I lament the fact that your presence is so often a subtractive force on the quality of the conversation instead of a substantiative and engaging one.

To address your comment: there are plenty of ways to disagree with MQA as an approach, technology, and product, but accusing Bob of not doing the homework is blasphemous. He's an AES Life Fellow and a cursory search shows him credited as an author nearly 20 AES papers.

 

[scott wurcer]

THE WHITE PAPER! BEHOLD! (Time stammped, just press play)

I love Thunderf00t

 

[filter_listener]

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.

Why don't you address the points in misewall's post to start?

You seem very focused on the "lossless" thing which I don't think anyone is really criticizing you about. Putting a PCM recording through the MQA encoder and playing it back will not result in a bit-perfect representation of the original PCM recording... therefore not "lossless" in the traditional sense. Not controversial. MQA has acknowledged this.

 

[voodooless]

therefore not "lossless" in the traditional sense. Not controversial. MQA has acknowledged this.

Funny that:

This question often seems to assume that lossless is always best but in fact all "lossless" does is to take some bits and to reproduce those same bits at another time or place. It that's all you wanted to do, FLAC would be fine and there would be no need for MQA.

We can stop right there! FLAC is all we need! I can stream high res audio from my phone over 4G.. I have multi-hundred megabit internet line at home, we watch 4K streaming video all the time without any problems. Why would I need MQA?

 

[scott wurcer]

I genuinely think it's sad that you value trolling and toxicity over discourse and education. Scoring internet points obviously brings you joy and who am I to say wether that's a good thing or a bad thing... but I lament the fact that your presence is so often a subtractive force on the quality of the conversation instead of a substantiative and engaging one.

You've done all your homework and read some or all or any of mansr's technical contributions? Otherwise this is a baseless insult. BTW IIRC when MQA was first introduced Mr. Stuart was asked if any controlled listening tests were done to support the claim of perceptually lossless and the answer was no, this is not science. Feel free to post links to any, I'd be happy to be wrong.

Almost forgot, can you put the meaning of this in your own words? BTW lossless compression is a lot more than 30 years old, the claim below is absurd.

Is it better than lossless? Yes, that's the sort of progress you should expect from the world-class team who developed lossless compression in the first place (30 years ago).

 

[mieswall]

Have you compared using a blind ABX test to eliminate confirmation & other bias? If no, then this is no different than claiming fancy cables sound better.

Sort of: I have a Mac Mini running Tidal app, and two DAC's by USB , a Chord 2Qute and a Project Pre Box S2. Amps are a truly amazing vintage Densen DM-10, or a EL-34 tubes push-pull class A integrated; speakers a pair of Kef LS50 Meta with a REL SW; Metas, btw, excel precisely in time coherence, imaging and soundstaging, and agility in transients, which may be a factor in my perception of these subjects, so important in MQA. I have had dozens of other stuff over the years, including much more expensive speakers, but I'm really happy on how this is sounding right now. Btw: until now, and with the remarkable exception of both the Chord and this Project under MQA, most DAC's I've tried for long periods of time (about a dozen, including very expensive ones), gave me little to null perceptual difference in truly A/B blind comparisons.

Sort of, because the Chord DAC sounds noticeably higher (3.0 V output instead of 2.02 of the Project), and so, some volume compensation is needed; I quickly figured out (confirmed by measurements with a Umik) the steps change in volume needed in the amp how to to minimize lag in the comparison (silence in between) to a couple of seconds. A key issue is to be sure to configure both the Mac and Tidal for the MQA rendering is to be fully unfolded by the DAC (this is NOT automatic, you must configure both Tidal passthrough and the Midi setup I the Mac; in fact it took me several tries until I got the blue light right when feeding it with MQA files).

To be honest, dealing with non-MQA content the Chord DAC still sounds a bit better (that's why I kept it in my system); but when it comes to true MQA there is simply no comparison. What's more, the higher the *true* sampling rate (not correctly seen in the computer but really in the DAC screen), the better it sounds. In some recordings it is just shocking.

Not long ago I listened in my system some amazing original masters tapes (recovered from a studio of Miami I understand) played with a spectacular Tascam R2R that a friend brought to my house. Up until then, it was the best sound I've heard in my place. MQA with good recordings, like recent ECM stuff, but specially 2L Audiophile Recordings files, are absolutely in the same level, emho.

I'm pretty sure that in better systems than mine which is rather modest, the difference with standard CD files would be even more evident.

 

[RichB]

Sort of, because the Chord DAC sounds noticeably higher (3.0 V output instead of 2.02 of the Project), and so, some volume compensation is needed; I quickly figured out (confirmed by measurements with a Umik) the steps change in volume needed in the amp how to to minimize lag in the comparison (silence in between) to a couple of seconds. A key issue is to be sure to configure both the Mac and Tidal for the MQA rendering is to be fully unfolded by the DAC (this is NOT automatic, you must configure both Tidal passthrough and the Midi setup I the Mac; in fact it took me several tries until I got the blue light right when feeding it with MQA files).

Volume matching is best done with a multimeter, not with a Umik mic.

To be honest, dealing with non-MQA content the Chord DAC still sounds a bit better (that's why I kept it in my system); but when it comes to true MQA there is simply no comparison. What's more, the higher the *true* sampling rate (not correctly seen in the computer but really in the DAC screen), the better it sounds. In some recordings it is just shocking.

I don't think you understand the meaning of "*true* sampling rate".
The only true sampling rate is that of the LPCM recording.
Source file up-sampling is a dime a dozen and royalty free

- Rich

 

[Tks]

Thanks for the tag! I should say first of all that I was just one of a group of folks over at ComputerAudiophile/AudiophileStyle who were deeply involved in the arguments over MQA, and the commentary I have made here about it has been deeply informed by the work @mansr and other folks did over there in the gigantic MQA thread.

That said, yes, I was quite intrigued to see @GoldenOne 's in-depth analysis, which built on mansr's and others' work. I don't think that detailed analysis Golden One did is irrelevant or redundant at this point: it provides empirical evidence of what we've been saying; it has some surprising new results that don't alter the basic issues but do further our understanding of this black-box technology; and it also contains a great one-stop list of informational links.

I don't think, however, that all this data is going to necessarily convince folks who don't want to be convinced. After all, there are more fundamental, basic, easier-to-understand principles of digital sampling theory that explain why, for example, very high sample rates are pointless, but that hasn't stopped the "audiophile=highest res possible" equation from taking root in both hardcore audiophile circles and in more casual hi-fi circles (for example vinyl ripper communities and high-res/HD streaming consumers). It is a great resource to point to, though, and is tremendously helpful because we can now just show some stuff is instead of having to try to explain how it logically must be the case.

As for @dmac6419 's comment that @amirm won't be touching this, that's probably true. Let me first say I deeply respect Amir's willingness to let folks (including me!) get into it with him so strongly about MQA. While this is a science- and fact-oriented forum, it's still his forum, and he's allowed folks to basically go off on him, and directly at him, about this topic more than most owner-mods of other forums would. That said, if one reviews Amir's comments in the forum here about MQA, I think one will find they do not quite fit his usual pattern of clearly and unsparingly condemning bad technological implementations and snake-oil BS, which is what MQA absolutely, positively is. But not every individual, even Amir, has to be equally interested in or passionate about critiquing every aspect of our hobby. When someone else posts a test of a piece of equipment, no one asks why Amir isn't testing that equipment himself. Same with MQA as far as I'm concerned - if he doesn't want to get into that fight, other folks can. It's a community, and as long as Amir allows the criticism to be made here, there's no reason he has to be compelled to join in if he doesn't want to.

Wasn't Amir there in that comments section in that other forum? Unless I'm just imagining, I could have sworn he was the one tearing people to pieces over MQA at that point in time, along with mansr, and I guess you as well. Also when I referenced someone debating with Amir on this forum about this perhaps shift on his views on MQA with respect to DRM, lol was that you? Bah I can't recall jack shit... But then again I never saw the conclusion to that discussion, nor can I really recall particulars of where and when that happened. I just remember asking him on clarification seeing as how his position seemingly changed, but I just took it to be out of some frustration over the criticisms levied against him for being in favor of DRM to the degree he was willing to perhaps at that junction even make a defense of MQA due to such reason (and not due to reasons of MQA itself being a good product or something). What I take to be the case is, he doesn't mind MQA since he has found perhaps variations of albums he's not found elsewhere. So even if MQA is garbage and still plagued with all the issues he riled it for all those years ago, it's nice to see MQA delivering perhaps on content otherwise not available from other publishers.

I hope this is what it is, and not that he believes MQA is something that delivers on it's marketing promises. But I ultimately don't care that much, I only cared to see what he thinks on this whole ordeal since I obviously value his opinion.

What I did care for, was hoping to see you and mansr see this post and comment, I'm satisfied that happened! Along with a few others.

 

[filter_listener]

We can stop right there! FLAC is all we need! I can stream high res audio from my phone over 4G.. I have multi-hundred megabit internet line at home, we watch 4K streaming video all the time without any problems. Why would I need MQA?

You don't need MQA! Nobody needs MQA! If you don't like it you shouldn't use it!

MQA is an attempt to:

A) Improve end-to-end time domain reproduction by utilizing b-spline convolution & resampling + complementary filtering(1)(2)(3)
B) Reduce the file-size required to store high-resolution audio(4)(5)
C) Provide provenance information for the file to verify that the audio has not been altered in transit (7)(8)

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

(1) https://www.soundonsound.com/techniques/mqa-time-domain-accuracy-digital-audio-quality
(2) https://patents.google.com/patent/US10867614B2/
(3) https://patents.google.com/patent/US20200395918A1
(4) https://patents.google.com/patent/US9548055B2
(5) https://bobtalks.co.uk/blog/mqaplayback/origami-and-the-last-mile/
(6) https://patents.google.com/patent/US20160005411A1
(7) https://bobtalks.co.uk/blog/science-mqa/16b-mqa-what-is-it/
(8) MQA allows for 24-bit -> 16-bit truncation by including signaling in the 16th bit which provides a lossy representation of the HF content

 

[sandymc]

MQA is an attempt to:

A) Improve end-to-end time domain reproduction by utilizing b-spline convolution & resampling + complementary filtering(1)(2)(3)
B) Reduce the file-size required to store high-resolution audio(4)(5)
C) Provide provenance information for the file to verify that the audio has not been altered in transit (7)(8)

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

 

[Tks]

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

Particularly B) and C) in totality. While A) is up in the air in principal due to semantics on what "time domain reproduction" improvements can be had over true lossless for example.