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

[Burning Sounds]

When I first bought it, I had a monthly subscription on Roon, for about 3 months, and I could see it in there, but I do not have that subscription any more.
When I downloaded the Wav-files, and converted them, I tested the resulting Flac-files with the MQA Tag Renaming Application
I currently do not have any means af playing the files as MQA, and probably never will.

I just tried to download that app - looks like it is not available for download now.

 

[jensgk]

I just tried to download that app - looks like it is not available for download now.

I downloaded it today, but you must fill out the form first.

 

[Burning Sounds]

I downloaded it today, but you must fill out the form first.

This is what I get - looks like a partial form and there is no box to check

 

[Chrispy]

I bought this album on Bandcamp: Kronos Quartet - Terry Riley- Sun Rings
I wanted the Flac version and downloaded it, it turns out it was 48/24 MQA files.
Since I would rather have pure 48/24 Flac files, I then downloaded the Wav files instead, thinking I could convert them to pure Flac.
So that is what I did. It then turns out that the converted Wav-files were also MQA, so the MQA was embedded in the Wav-files.
In this case I was not warned about getting MQA, and could not avoid it beforehand.
Sure, I expected 44.1/16 files, but got 48/24 MQA instead, but I really do not want MQA...

Curious, did you let Bandcamp and the artist know of your preferences?

 

[mieswall]

@amirm, perhaps you can answer some simple questions.

• Is it possible for an MQA 44.1 kHz 16 bit file to deliver CD quality Redbook audio?
• Is it possible for a user to decode MQA to obtain the best dynamic range and choose your an alternate reconstruction filter?
• Has anyone been able to prove that Temporal Blur is a problem and the MQA solution is audible, say in a blind test?
• What exactly does the consumer, with bandwidth enough to stream Netflix, get out MQA?

- Rich

Isn't a bit of a contradiction to ask what temporal deblurring achieves, but at the same time, be so upset with MQA because a bit-perfect match was not found in these test? What would happen if somebody demonstrates there are audible effects in that deblurring? Still the tests would be prominent in your judgment?

In turn you, but ideally GoldenEar may answer:

• Is it right to ask for a bit-perfect match to a system that by design replaces floor noise with information? If not, wouldn't that be enough to invalidate the whole test that pretends to qualify the system by this criteria? And please don't argue that this is hidden information, because this is explained in every single article about MQA, the issue is so well known that even without reading any article of them you would know it in advance. If there is one thing for sure about MQA, is that. Or do you believe the folding process just makes bit space for the folds to appear out of nowhere?
  
  
• If the MQA patent refers to previous art, patents of their team specifically dealing with noise shaping, a research done by this team literally for decades, why do you think those references are for, when those patents and research deal exactly with the way to make use of that otherwise useless and inaudible noise space? Then, even knowing that, why do you still will keep measuring as if that useless original noise is to be kept intact? And to add insult to the subject, then make a scandal because the bitmap doesn't look the same, which is this way by definition?
  
  
• If it is accepted, as it is implicit in the question of RichB, that MQA tries to correct time domain issues (leaving aside for now your doubt about how useful or not it is; if someone doesn't bother to read the articles explaining that, he is in his right to keep in doubt and ask, but not condemn until acquiring that knowledge); therefore with the implicit consequence that the source file will be modified... Is it right to ask again for a bit-perfect match which is precisely what the system is trying NOT to achieve?
  
  
• If the system is looking for a noise floor to process it, as every recording has by definition as analyzed in unbelievable extents by MQA, and you provide a signal without it because you didn't add a dithered noise to the file... are you expecting that system to behave the same way, as you seems to be so alarmed that the ground noise is showing anomalies, aliasing, and the flaws that are precisely the things this design is tying to fix?
  
  
• If documents of MQA go in great extents about the dynamic envelope of music, pages and pages about it, dozens of graphs with luxury of details, and btw a fact backed up by the very physics of music, why do you think they go into those extents? If you conclude in fact that they are trying to make use of that analysis instead of just wasting ink, what else instead of discarding those higher amplitudes without music could this discussion be aimed for?
  If the unavoidable conclusion is that, in fact, those upper band and ultrasonic high amplitudes will be discarded in MQA, why are you measuring content that you know in advance that the system is not designed to deal with? Are in you opinion those ultrasonic high amplitudes of square waves or white noise useful for the music the system is trying to faithfully with?; if the documentation informs you advance this analysis of music content is because their intention is to recover that space for other uses (filters are specifically mentioned)?, Then, is it appropriate to conclude the system if behaving badly to those tests you were previously informed that they wouldn't considered in the system? I know... because they used the forbidden word: they say they are "lossless", and that is egregious to you if they can't fully reconstruct a square wave.... But the point here is they say they are lossless to the music content in the master they are using as input, not against middle steps in the process or any possible content regardless if it is music or not, which is what you are measuring here.
  
  
• In sum, did you read any paper or article about MQA design principles before doing those tests? This is a tough question: If you say yes, then why you prepared a test that purposely contradicts those documents? Is you say no, well...no further comments other than the mention of amateurism is precisely because of this.

We all may disagree with those MQA axioms, we may even find it is all vaporware, and those axioms would be a useful thing to discuss here. But what's wrong is to justify this credo with a test that is deeply, fundamentally bad designed.

I was eager to read GoldenEar answer to these issues, sadly he didn't.
To be fair, I think he has a point about how closed the details of MQA are and unhelpful MQA could be about testing its system. I do agree with him in that. But that doesn't justify to make a faulty test (in fact, exactly the opposite: if you don't exactly understand what the system is doing, the responsible thing to do is to inform yourself first; and please don't argue that they are hiding this: I am much less knowledgeable that you and yet, I am aware of the things I'm saying, just reading the information that is available in the internet).

Either way, a wrong test, driven by naive enthusiasm could be a mistake. What is not a mistake is to drive the kind of conclusions GoldenEar did about those tests, because he knew in advance the kind of results he would get feeding the test with out-of-context data. It is at least irresponsible. And it is more serious, as the kind of accusations in the conclusions of the video are just insulting. You may not agree with the business model of MQA, but using a purposely wrong test to demonstrate ... what? and then tell the world that MQA people are a bunch of con-artists is completely disgusting, in my opinion.

That's why GoldenEar there are some here upset with this. You are purposely leaving a stain in what I believe, and my ears tell me, is a system that performs admirably playing music (If the criteria is how it sounds, installing a MQA DAC is among the best upgrades I have ever made to my equipment, for a fraction of other costs) Which in the end, I thought it was the purpose of this, after all. The reason why we all read this site.

 

[jensgk]

This is what I get - looks like a partial form and there is no box to check

You need to fill out the fields, and tick off the boxes, - then the Download button is activated

 

[jensgk]

Curious, did you let Bandcamp and the artist know of your preferences?

No, I did not, but maybe I should.

 

[RichB]

Isn't a bit of a contradiction to ask what temporal deblurring achieves, but at the same time, be so upset with MQA because a bit-perfect match was not found in these test? What would happen if somebody demonstrated there are audible effects in that deblurring? Still the tests would be prominent in your judgment?

In turn you, but ideally GoldenEar may answer:

• Is it right to ask for a bit-perfect match to a system that by design replaces floor noise with information? If not, wouldn't that enough to invalidate the whole test that pretends to qualify the system by this criteria? And please don't argue that this is hidden information, because this is explained in every single article about MQA, the issue is so well known that even without reading any article of them you would know it in advance. If there is one thing for sure about MQA, is that. Or do you believe the folding process just makes bit space for the folds to appear out of nowhere?
  
  
• If the MQA patent refers to previous art, patents of their team specifically dealing with noise shaping, a research done by this team literally for decades, why do you think those references are for, when those patents and research deal exactly with the way to make use of that otherwise useless and inaudible noise space? Then, even knowing that, while do you still will keep measuring as if that useless original noise is to be kept intact? And to add insult to the subject, then make a scandal because the bitmap doesn't look the same, which is this way by definition?
  
  
• If it is accepted, as it is implicit in the question of RichB, that MQA tries to correct time domain issues (leaving aside for now your doubt about how useful or not it is; if someone doesn't bother to read the articles explaining that, he is in his right to keep in doubt and ask, but not condemn until acquiring that knowledge); therefore with the implicit consequence that the source file will be modified... Is it right to ask again for a bit-perfect match which is precisely what the system is trying NOT to achieve?
  
  
• If the system is looking for a noise floor to process it, as every recording has by definition as analyzed in unbelievable extents by MQA, and you provide a signal without it because you didn't add a dithered noise to the file... are you expecting that system to behave the same way, as you seems to be so alarmed that the ground noise is showing anomalies, aliasing, and the flaws that are precisely the things this design is tying to fix?
  
  
• If documents of MQA go in great extents about the dynamic envelope of music, pages and pages about it, dozens of graphs with luxury of details, and btw a fact backed up by the very physics of music, why do you think they go into those extents? If you conclude in fact that they are trying to make use of that analysis instead of just wasting ink, what else instead of discarding those higher amplitudes without music could this discussion be aimed for?
  If the unavoidable conclusion is that, in fact, those upper band and ultrasonic high amplitudes will be discarded in MQA, why are you measuring content that you know in advance that the system is not designed to deal with? Are in you opinion those ultrasonic high amplitudes of square waves or white noise useful for the music the system is trying to faithfully with?; if the documentation informs you advance this analysis of music content is because their intention is to recover that space for other uses (filters are specifically mentioned)?, Then, is it appropriate to conclude the system if behaving badly to those tests you were previously informed that they wouldn't considered in the system? I know... because they used the forbidden word: they say they are "lossless", and that is egregious to you if they can't fully reconstruct a square wave.... But the point here is they say they are lossless to the music content in the master they are using as input, not against middle steps in the process or any possible content regardless if it is music or not, which is what you are measuring here.
  
  
• In sum, did you read any paper or article about MQA design principles before doing those tests? This is a tough question: If you say yes, then why you prepared a test that purposely contradicts those documents? Is you say no, well...no further comments other than the mention of amateurism is precisely because of this.

We all may disagree with those MQA axioms, we may even find it is all vaporware, and those axioms would be a useful thing to discuss here. But what's wrong is to justify this credo with a test that is deeply, fundamentally bad designed.

I was eager to read GoldenEar answer to these issues, sadly he didn't.
To be fair, I think he has a point about how close and unhelpful MQA could be about testing its system. I do agree with him in that. But that doesn't justify to make a faulty test (in fact, exactly the opposite: if you don't exactly understand what the system is doing, the responsible thing to do is to inform yourself first; and please don't argue that they are hiding this: I am much less knowledgeable that you and yet, I am aware of the things I'm saying, just reading the information that is available in the internet).

Either way, a wrong test, driven by naive enthusiasm could be a mistake. What is not a mistake is to drive the kind of conclusions GoldenEar did about those tests, because he knew in advance the kind of results he would get because he knew he was feeding the test with out-of-context data. It is at least irresponsible. And it is more serious, as the kind of accusations in the conclusions of the video are just insulting. You may not agree with the business model of MQA, but using a purposely wrong test to demonstrate ... what? and then tell the world that MQA people are a bunch of con-artists is completely disgusting, in my opinion.

That's why GoldenEar there are some here upset with this. You are purposely leaving a stain in what I believe, and my ears tell me, is a system that performs admirably playing music (If the criteria is how it sounds, installing a MQA DAC is among the best upgrades I have ever made to my equipment, for a fraction of other upgrades) Which in the end, I thought it was the purpose of this, after all. The reason why we all read this site.

Admittedly, I scanned because it became clear that you would not be directly addressing questions.
Here is a cheat sheet in case you try again:

• Is it possible for an MQA 44.1 kHz 16 bit file to deliver CD quality Redbook audio?
  No, it is not MQA destroys those bit binary values.
• Is it possible for a user to decode MQA to obtain the best dynamic range and choose your an alternate reconstruction filter?
  I don't think so, MQA requires their reconstruction filter.
• Has anyone been able to prove that Temporal Blur is a problem and the MQA solution is audible, say in a blind test?
  None that I have seen, SBT tests have not found a significantly audible difference.
• What exactly does the consumer, with bandwidth enough to stream Netflix, get out of MQA?
  There is no benefit only potential costs

This is absolutely not the site where people blindly accept what peoples ears tell them.

Good luck with your ear stain.

- Rich

 

[Chrispy]

No, I did not, but maybe I should.

Well there is the voting with your $ thing. Thanks for the headsup, tho, I wouldn't expect the mqa pollution on a download unless it says so.

 

[jensgk]

Well there is the voting with your $ thing. Thanks for the headsup, tho, I wouldn't expect the mqa pollution on a download unless it says so.

You are right. I just sent them a note.

 

[Burning Sounds]

You need to fill out the fields, and tick off the boxes, - then the Download button is activated

I filled out the fields (there is only a First Name field and email address field), there are no boxes to tick off, so nothing gets activated. I've tried with Firefox, Chrome and Edge.
Strange - seems to work on my phone, though. MQA are sailing close to the wind regarding GDPR as it insists on consenting to their email updates to be able to sign up for the download. Not good practice.

 

[Ralph_Cramden]

OK, I'm confused. This site is supposed to be all about measuring stuff, to see whether it's worth buying. @GoldenOne made a pretty decent attempt to measure MQA, and some folks here appear to be freaked out. WTF!?

 

[AdamG]

Notice to all Thread participants: This tread has caused significant and unending arguments and circular logic. Beating a dead horse comes to mind. So, it is my decision to close this thread to further comment tomorrow afternoon. Get in your last words now. Please try to be brief

 

[Cahudson42]

Well, my take. MQA is a perfect example of a 'solution looking for a problem'.

And the -' problem' is not audio quality but maximum revenue extraction.
'Bye!

 

[mkawa]

are any of you paying for mqa? honest question. answer truthfully and without discussion about technical details. i am curious, because i've paid a lot for a lot of equipment but i don't remember paying anything extra for MQA, either access to it, or streaming subscriptions that included it as part of access to RB

 

[muslhead]

dammit, now what am i going to do tomorrow? watch soap operas?

 

[muslhead]

are any of you paying for mqa? honest question. answer truthfully and without discussion about technical details. i am curious, because i've paid a lot for a lot of equipment but i don't remember paying anything extra for MQA, either access to it, or streaming subscriptions that included it as part of access to RB

topping dac. mqa version is $50 more, i believe

 

[Cahudson42]

are any of you paying for mqa?

No. But an issue is it's worming it's way into what I do pay for - 44.1/16 FLAC and compromising it.

 

[scott wurcer]

dammit, now what am i going to do tomorrow? watch soap operas?

I'll go back to binging "Great British Menus"

 

[RichB]

I will continue to hope that @Amir is right and MQA dies on the vine.

Dishonor goes to Tidal for obfuscating their offerings replacing lossless formats with NQA with nebulous terms like Hi Fi and Master.

Kudus to QOBUZ for providing clear streaming plans that are true lossless, true masters, unrestricted, in standard and open CD and Hi-Res audio formats.

I don't recall but do Hi-Res audio threads require forced closure?
If not, it must be MQA that is the problem.

- Rich