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

[JSmith]

Is it a problem to participate in more than one forum?

Certainly not my friend... I am not only here myself. However always a good idea not to piss in ones pond, which the OP can't seem to grasp going by comments made on other forums. As we all know many forums are interconnected on subject, so best to keep consistent from forum to forum and not play one off against the other for notoriety.



JSmith

 

[DimitryZ]

Your wording in the section I quoted (ignoring the spelling error) expresses support for the veracity of those accusations, but I suspect that is not what you meant. Hence the request for clarification.

I guess I don't understand?

@mansr has been openly accusing this forum's owner of being on MQA payroll (on PFM) - but now it has been cleaned up after my request.

He has previously leveled the same accusation at me.

Apparently, he believes that MQA runs a special closed Facebook page, where people like me and others congregate and get instructions from MQA with "talking points" so we can spread out on audio forums and desiminate MQA propaganda.

 

[andymok]

Well...

while I remain convinced that MQA is a pointless money grab for the following reasons

I rmb it was the time when streaming services got popular, mobile data was still slow and limited, and there was a demand for hi-res where services can charge a premium (pls correct me if i'm wrong)

so I always thought, MQA is a way to encourage ppl to pay more, by boosting consumer's confidence with "Authentic", whilst services can charge more and musicians/agencies earning more, MQA on the other hand sells hardwares, IPs and licenses etc....... A win-win-win-win solution.

so it always seems to me it's about commercials, way more than technicalities, consumer benefits whatsoever
yeh yeh it's not charity I know ......

 

[ebslo]

I guess I don't understand?

It's possible I am wrong and you meant just what you said, but in any case, you may want to check the definition of "scrupulous" just to make sure.

 

[Raindog123]

Whilst I have some sympathy for that position, nevertheless
1) end to end impulse response was one of the heavily touted features of MQA
2) in the frequency domain the triangle of music is supposed to include at least potentially stuff over 48kHz
3) IIRC some of the early promotional material suggested that there was soemthing like a second unfold (I seem to remember
a diagram with a zone C in the buried data), although I am aware that many people have asserted that the rendering is just upsampling (and even here there is the possibility of encoding through what Jim lesurf has called the leaky decimation filter)

....so on any basis it does not appear to be possible to analyse MQA as it claims to be without looking at the second unfold or rendering stage. The idea that MQA was just a clever way of packing 24/96 seems to be something of a novelty.

All good points, worth some discussion. Unfortunately, any meaningful conversation in this environment will get crapped on by those 'newcomers':

1) "End to end impulse response was one of the heavily touted features of MQA" - yes, the end-to-end signal-path correction with a reconfigurable digital filter is a viable tool. Is used by yours truly all the time in RF Tx-Rx chains... However it is a true point-to-point - implies having control of the encoder’ data at the source. That might not be available to MQA encoding in modern mastering, eg due to data multi-source, where the point-to-point no longer holds. [This is why we need mastering experts' wisdom here.]

2) "In the frequency domain the triangle of music is supposed to include at least potentially stuff over 48kHz" - I am not arguing that the '[music] triangle" does not go above 48kHz... But why would we care, with the hearing limit like this (the black, triangled curve) [from Meridien paper]:

3) "...and even here there is the possibility of encoding through what Jim lesurf has called the leaky decimation filter" - there is always a possibility for encoding. The issue there - and it too has been pointed out a lot - your baseband (especially if it's a 16-bit one) simply runs out information capacity to store any this-particular-tune related information - first the (<22kHz) baseband itself, then the 'first', and then this 'second' unfold. No matter what codec and how 'leaky' your filters, you only carry information (a) in your filter shape (a constant) convoluted with (b) some information ultimately taken from the initially-baseband samples. So, yes, the information budget of the 16bit/22kHz or 24bit/22kHz baseband needs to be looked at (and that's what the 'triangle' does, but unfortunately not quantitatively, plus see #2.)

"....so on any basis it does not appear to be possible to analyse MQA as it claims to be without looking at the second unfold or rendering stage." -- And this is why I posted my question of how to tap into the post-FIR of existing audio DACs, or host the exact MQA rendering on a more capable DSP HW (or in Matlab/Simulink for that matter), and capture the datasteam for post-analysis from there. We shall see...

 

[firedog]

Does this mean the number of (PC-)software full unfolders is zero? That'd really be dissapointing. So tapping the output of a firmware MQA final renderer chip like some XMOS is the only way?

Yes, and that's one of the issues with MQA.

 

[DimitryZ]

It's possible I am wrong and you meant just what you said, but in any case, you may want to check the definition of "scrupulous" just to make sure.

I meant scurrilous, of course.

 

[UliBru]

@Raindog123: Please be aware that talking about the "triangle" accompanied by pictures MUST show up the frequency axis in linear setting and not logarithmic!

 

[Inner Space]

It's possible I am wrong and you meant just what you said, but in any case, you may want to check the definition of "scrupulous" just to make sure.

I think he mean "scurrilous" ... second language problems, I guess. He can't punctuate, either, but hey.

 

[Raindog123]

Hey MODs: I know you're listening...

Per my earlier suggestion to you guys, why would not we split this MQA crap of a thread into two: a technical/scientific one... and the fingerpointing marketing BS conspiracies one!? And keep the two (and the respective actors) a mile away from each other.

Getting unbearably smelly here...

 

[earlevel]

You partially answer your own question there. There are two misunderstandings here:

1. That the encoder needs to handle all cases perfectly. It does not...

Sorry, good post, but I'm not misunderstanding. I implied the above in my posts, I thought.

And I didn't talk about levels, the posts I replied to didn't talk about levels. You didn't say square waves and white noise would be bad at full code (for instance), you said they were unrealistic. I simply said they were realistic. I wasn't trying to say you were wrong, I just was making a clarifying point about such signals. I even said that you were obviously talking about making inferences from measurements that weren't appropriate and not exactly what I'm talking about.

It's fine if an encoder doesn't handle extreme and unlikely cases. I'd hope that they would actually be extreme and unlikely, of course. If full-code white noise doesn't work out, that's fine. If white noise at -35 dB doesn't work out, in a noticeably bad way, something's wrong. If I know going in that MQA is particularly bad at these signals (I'm not saying it is—it probably isn't), of course I might want to test where that happens so I'm not always wondering if I heard a glitch when playing electronic music.

PS—From Bob Stuart's comments: "The blogger’s test failed because he submitted signals that do not resemble music to an encoder that was configured only for music works. Nonsense comes out." I'm sure Bob is exaggerating to some degree to make a point. But since some of these signals are not that far from music—square waves, white noise—I'd want to know more about the threshold and degree to which the output turns to "nonsense", based on the input. If he'd said something like it not producing good test results, OK, but "nonsense" is a pretty strong word.

 

[amirm]

I noticed the editors put a serious disclaimer on the MQA article

Editor's Note: Most of Bob Stuart's answers have been debunked and the MQA technology is now seen as lacking any benefit for anyone other than record labels and MQA Ltd.

I’m going to switch from Tidal. Amazon and Apple are streaming lossless to my DAC for free.

??? 95% of what I play on Tidal is NOT MQA. MQA makes up a tiny fraction of content I find on Tidal. Most of the time I have to seek it out to find it. Please don't run with folks who have other reasons to hate MQA. Chris who wrote that is making a business out of anti-MQA. He is not technical and doesn't understand any of these topics himself.

 

[amirm]

PS—From Bob Stuart's comments: "The blogger’s test failed because he submitted signals that do not resemble music to an encoder that was configured only for music works. Nonsense comes out." I'm sure Bob is exaggerating to some degree to make a point. But since some of these signals are not that far from music—square waves, white noise—I'd want to know more about the threshold and degree to which the output turns to "nonsense", based on the input. If he'd said something like it not producing good test results, OK, but "nonsense" is a pretty strong word.

They are *very* far from music. No one in this thread or elsewhere has produced a single piece of music that is similar to spectrum of those test tones or even close. The last sample was produced that sounded noise like still followed an exponential drop off and is unlikely to have much in ultrasonic range:

Remember, you can NOT play OP's test signal on your system without potentially damaging it. Your tweeter, nor your ears can handle high amplitude high frequencies. And there would be no reason for an artist to generate artificial sounds that have ultrasonic content let alone extreme amount of it. All music is listened to by humans which guarantees you that kind of drop off. I can't imagine anyone releasing content that instantly blows up the tweeter in your system, causes the amp to oscillate and you running away with a scream!

Sure, can someone do it on purpose like OP did as a prank? Yes. In that case what MQA does to it doesn't matter. The content itself is not high fidelity and no one would care one bit what the noise level underneath all of that "signal" would be.

 

[PierreV]

I rmb it was the time when streaming services got popular, mobile data was still slow and limited, and there was a demand for hi-res where services can charge a premium (pls correct me if i'm wrong)

so I always thought, MQA is a way to encourage ppl to pay more, by boosting consumer's confidence with "Authentic", whilst services can charge more and musicians/agencies earning more, MQA on the other hand sells hardwares, IPs and licenses etc....... A win-win-win-win solution.

so it always seems to me it's about commercials, way more than technicalities, consumer benefits whatsoever
yeh yeh it's not charity I know ......

Yes, yes. Starting with potentially very good ideas about lossy codecs when they considered the bandwidth problem, then finding a way to repurpose the work as a "luxury" alternative, which is the area of expertise of the Richemont group.

 

[amirm]

I dont care about MQA at all but the side discussion about what can be considered music is interesting to me. I would like your opinion on the possibility of encoding some more experimental music.

A poster above already mentioned Ryoji Ikeda. A japanese electronic music artist that is known for using sine waves at the end of the audible spectrum and random noise. I actually enjoy listening to this on occasion.

MQA would have no trouble with such content. It can just pass it through as is, while extending its bit depth to 24 bit and apply noise shaping/subtractive dither to it for better dynamic range. It has no obligation to encode ultrasonics that may not be there and can easily throw them out if there is any based on input from the operator.

Optionally, you can refuse to encode such a file. I run into content on Tidal occasionally that is only in AAC format, it is not even lossless CD. I listen to it and it is fine versus not having that music. By the same token, the operator if they cared about effects of that high-res data, could release it only in CD format, or flac lossless. It is highly doubtful though.

Circling back, the baseline layer in MQA is 24-bit 44.1/48 kHz. This is *massive* bandwidth that it is allowed to generate even in worst case scenario and can do a better job of containing high-res/high-dynamic range data than 16/44.1 can.

 

[q3cpma]

Chris who wrote that is making a business out of anti-MQA.

How would someone even make money this way? I have to be honest, you use the expression "conspirary theory" with disdain, but you yourself allude to an organized anti-MQA camp with possibly paid shills within it, please provide a modicum of proof.

 

[amirm]

On the other hand, for those who have a MQA-DAC, ROON is the only way to use DSP and at the same time utilise the MQA capabilities in the DAC. That is also a limitation.

What limitation? As I have explained multiple times, there are only two mainstream players that consume Tidal content with MQA and one is Tidal which doesn't have DSP, and the other is Roon that does. If you choose to use the Tidal app, then my sympathies are in order. I only use it in my car. For the rest, you best use Roon and there, you have your total solution. If you don't want to use either, then you are not a Tidal customer, so would not get MQA content anyway so the objection is moot.

 

[adamd]

All good points, worth some discussion. Unfortunately, any meaningful conversation in this environment will get crapped on by those 'newcomers':

2) "In the frequency domain the triangle of music is supposed to include at least potentially stuff over 48kHz" - I am not arguing that the '[music] triangle" does not go above 48kHz... But why would we care, with the hearing limit like this (the black, triangled curve) [from Meridien paper]:
View attachment 133170

3) "...and even here there is the possibility of encoding through what Jim lesurf has called the leaky decimation filter" - there is always a possibility for encoding. The issue there - and it too has been pointed out a lot - your baseband (especially if it's a 16-bit one) simply runs out information capacity to store any this-particular-tune related information - first the (<22kHz) baseband itself, then the 'first', and then this 'second' unfold. No matter what codec and how 'leaky' your filters, you only carry information (a) in your filter shape (a constant) convoluted with (b) some information ultimately taken from the initially-baseband samples. So, yes, the information budget of the 16bit/22kHz or 24bit/22kHz baseband needs to be looked at (and that's what the 'triangle' does, but unfortunately not quantitatively, plus see #2.)
.

Just checking we are seeing eye to eye here- I'm not making any claim about audibility of >48kHz information here, just wishing to point out that this idea that sort-of lossless 24/96 compression does not sit well with the other design aims of MQA which I understand to require the encoding of >48 kHz material in the 24 (perhaps 22Khz) -48Khz material (this is already part of the information encoded in the <24Khz undecoded file.) My understanding of how this might work (assuming that there is only one unfold) would require MQA to be lossy in the 24-48khz region on any view ie it would be incompatible with the sort-of-lossless claim. It stands apart from the question of how much room there is to encode the 24-48Khz material in the undecoded file.

 

[amirm]

How would someone even make money this way?

Advertising on his site. He has made this his cause célèbre and benefits hugely from all the hits as that is his main source of income and livelihood. Outside of MQA, Chris is a huge subjectivist with no regard or care whatsoever about proper audio science and measurements. Indeed for a while he completely banned objectivity talk from his forum and banned a bunch of members which is the reason we have people like Mansr here, rather than there. There, they rolled out a red carpet for Mansr to complain about MQA. But outside of that, Chris realized it was not good for sponsorship and advertising business so he and many others were shown the door.

OP is in the same situation by the way with his youtube revenues and lack of any objectivity in audio prior to this experiment.

I find it very dishonorable to take position against other companies on such basis when in the rest of their audio lives, they could care less about measurements, objectivity, etc.. In the morning they are vegetarian but at dinner they eat steak while telling people eating meat is bad.....

 

[PierreV]

Please don't run with folks who have other reasons to hate MQA. Chris who wrote that is making a business out of anti-MQA. He is not technical and doesn't understand any of these topics himself.

Dr Sean Olive essentially said the same thing a few days ago. Did you miss it?
Is Dr Sean Olive not technical enough? Is he involved in that mysterious anti-MQA crusade as well?
Or maybe simply a "tool" as you qualified those of us who weren't enthusiastic about MQA earlier?