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

[amirm]

I will answer it for you

Well, I‘ve just counted, and I probably missed a few...

I personally asked about “benefits of the 16/44 MQA (aka MQA-CD)” within this thread 7 times! Got 0 (zero) answers.

The content owner can advertise that with the suitable decoder, the same bits will become high-resolution -- something the CD can't do.

I asked for a “proof of [any] MQA sounding better than non-MQA, through controlled listening or measurements” 11 times! Guess what, 0 (zero) responses.

MQA team produced an award winning peer reviewed paper that showed filtering high res audio can create audible differences. This lay the foundation for MQA not peforming such down sampling. They did not go past this to produce other studies. The anti-MQA team has also not produced any listening test to show that application of MQA to CD format is harmful.

I asked about why we need ultrasonics at least 5 times (while the hearing curve published by MQA/Meridian stops under 20kHz)... same story.

I must be a cellophane man... Or, maybe there are no answers!?

MQA is not just about high sample rates but also bit depth. I have shown a lot of evidence as to why we need this. I have also shown that I have passed one such blind test. The extra bandwidth is also useful for noise shaping and can avoid aliasing artifacts with some non-linear effects. Overall the benefits are small but to the extent they are there, why not.

Note that the market demand for high-res is much higher than what objectively can be explained. MQA is taking advantage of that need to to deliver a solution.

 

[DimitryZ]

As tonight is the night of demystifying everything at once, can you have a shot at my three long-overdue questions? Here they are:

MQA Deep Dive - I published music on tidal to test MQA | Page 93 | Audio Science Review (ASR) Forum

Hopefully, they will help us to knock-off that final one "Why do consumers need MQA (both 24/48 and 'MQA-CD') if we already have 24/48 (and can have 20/96) open-standard PCM/FLAC"?

Thanks in advance, @DimitryZ. Or should we call you 'Bob'? (Sorry, Twin Peaks is still my all-time favorite!)

Absolutely. Though your professional credentials seem ever thinner as you post.
MQA simply sounds better than LPCM counterpart.
Now go back to your golden parachute and enjoy your retirement.

 

[earlevel]

All music falls into the same statistical envelope. People naturally assume that strange music (to their ears) like "noise" genre or beautiful gamelan must be different. It is not.

You don't specify, but it seems like you mean spectral envelope. If so, you're overstating it with "all". It's true that most music that most people like, tends to fall in a similar spectral envelope (broadly speaking, "pinkish"—and natural instruments have harmonics that drop off relatively quickly), but even that varies by genre to some degree. Here's the music I posted above, compared with a typical classical spectral envelope:

If I misinterpreted what you meant, please let me know. I basically agree as a broad generalization, but "all music falls into the same statistical spectral envelope" is overstating it, unless the envelope is so wide as to not be useful.

 

[DimitryZ]

You don't specify, but it seems like you mean spectral envelope. If so, you're overstating it with "all". It's true that most music that most people like, tends to fall in a similar spectral envelope (broadly speaking, "pinkish"—and natural instruments have harmonics that drop off relatively quickly), but even that varies by genre to some degree. Here's the music I posted above, compared with a typical classical spectral envelope:

View attachment 133075

If I misinterpreted what you meant, please let me know. I basically agree as a broad generalization, but "all music falls into the same statistical spectral envelope" is overstating it, unless the envelope is so wide as to not be useful.

You have misitreprented what I stated .

If your example exceeds MQA published envelope, please show us where

Otherwise, don't bother.

 

[KSTR]

https://bobtalks.co.uk/a-deeper-look/all-that-glitters-is-not-golden/#

Sadly it does not contain the info I'm looking for.
Their "corpus peak level plot" shown in https://bobtalks.co.uk/a-deeper-look/appendix-2-test-signals-and-music/ is no help either. I've shown a snippet from a random CD where the HF level would heavily exceed those gathered limits, no FFT needed, the waveform already tells everthing.
How to solve this conundrum?

 

[JSmith]

JSmith

 

[earlevel]

You have misitreprented what I stated .

If your example exceeds MQA published envelope, please show us where

Otherwise, don't bother.

Why be so rude? No point was missed—if you overstated, fine, but don't tell me what I need to do. You said, in reply to me, that all music fits the same "statistical envleope". Clearly, this is not true unless the envelope is absurdly broad. Any "high fidelity" encoder should allow signals that aren't your expectation of a "statistical" musical envelope.

I gave an example of unexpected music, not white noise, it was merely a related comment that we shouldn't have too narrow a view of what qualifies as music. I don't need to check with MQA, it wasn't about MQA, it was about Amir's comment of suitable source material in testing. White noise and square waves should absolutely be expected in audio that would end up in a streaming service. Music or otherwise—Apple Music has plenty of sleepy-time noise tracks. The solo in Lucky Man is unfiltered square waves (Emerson just got his modular, hadn't yet learned the joys of filtering—sure, it's a chorus of square wave oscillators, plus drums, but I'm sure I could fine you plain square waves in Kraftwerk or somewhere else, white noise bursts too). I think Amir might have meant their use beyond listening tests, as I allowed in my comment to him.

 

[jensgk]

..
Do you see square wave here? No.
Do you see impulse? No.
...

Do you see 1khz test tones here? No.

 

[restorer-john]

Music or otherwise—Apple Music has plenty of sleepy-time noise tracks. The solo in Lucky Man is unfiltered square waves (Emerson just got his modular, hadn't yet learned the joys of filtering—sure, it's a chorus of square wave oscillators, plus drums, but I'm sure I could fine you plain square waves in Kraftwerk or somewhere else, white noise bursts too). I think Amir might have meant their use beyond listening tests, as I allowed in my comment to him.

Have a listen to early Jean Michel Jarre, he's got a bit of everything in there.

 

[Raindog123]

I will answer it for you

Thanks BOB, I mean, Amir.

I personally asked about “benefits of the 16/44 MQA (aka MQA-CD)"...

The content owner can advertise that with the suitable decoder, the same bits will become high-resolution -- something the CD can't do.

Sorry, I do not understand that. What 'the same bits become high-resolution'? Can you please elaborate?

I asked for a “proof of [any] MQA sounding better than non-MQA, through controlled listening or measurements”...

MQA team produced an award winning peer reviewed paper that showed filtering high res audio can create audible differences. This lay the foundation for MQA not peforming such down sampling. They did not go past this to produce other studies. The anti-MQA has also not produced any listening test to show that application of MQA to CD format is harmful.

I got the "audible differences" part... But what about "sounding better"? I think we've discussed - in this very thread - the listeners in the paper hearing a difference between interpolation filters does not point at any benefit. Any up-sampling can use any convolutional filter they wish, and many of then will sound 'audibly different'?

The gist of my question/point is not ' that application of MQA to CD format is harmful' but rather is why use a proprietary, closed, untested codec if it is not beneficial?

Finally, as you kindly pointed out yourself, the MQA guys themselves produced this evidence (see quoted figures 10 and 11) of the hearing threshold being limited by 20kHz (the thick solid lines)...

I asked about why we need ultrasonics at least 5 times (while the hearing curve published by MQA/Meridian stops under 20kHz)...

MQA is not just about high sample rates but also bit depth. I have shown a lot of evidence as to why we need this...

So, from the bit-depth point of view, the 24/48 PCM is superior to 24/48 MQA!? As MQA's bit-depth is lower than 24-bit - due to their need to encode the ultrasonics in them. I think we're in agreement there!

So, what about my bonus question, to leave nothing unturned:

Why do consumers need MQA (both 24/48 and especially 16/48 'MQA-CD') if we already have 24/48 (and can have 20/96) open-standard PCM/FLAC"?

 

[jensgk]

Archimago's analyses have been publically available for four years - plenty of time for anti-MQA folks to get educated, instead of yucking it up on various forums. For the consumer, his analysis tells you how it works what its' flaws and weaknesses are.

There are lots of unknowns. Without further testing, breaking and reverse engineering, it is simply not possible to make independent software, that bit perfect can encode and decode MQA.

 

[jensgk]

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

Exactly, it does not. It is broken.

After all, a good lossless codec would detect what is "below the trangle", based on the actual, individual piece of PCM sample, and just encode what is actually in that sample.

If there are any violations, it would be handled gracefully, and encoded losslessly, maybe not compressed anymore, but it would be done.

I can be very foregivefull for lossy codecs, I do not expect them to behave very well. But a codec, that again and again claims some form of losslessnes should not be so easy to break, and then actually make a file, that is playable. That is bad engineering, and cannot be trusted.

 

[JohnYang1997]

Exactly, it does not. It is broken.

After all, a good lossless codec would detect what is "below the trangle", based on the actual, individual piece of PCM sample, and just encode what is actually in that sample.

If there are any violations, it would be handled gracefully, and encoded losslessly, maybe not compressed anymore, but it would be done.

I can be very foregivefull for lossy codecs, I do not expect them to behave very well. But a codec, that again and again claims some for of losslessnes should not be so easy to break, and then actually make a file, that is playable. That is bad engineering.

MQA is not lossless hence everything you said afterwards doesn't have a meaning.
It should be by now a known fact about MQA.
You missed what you are arguing for.

 

[earlevel]

Have a listen to early Jean Michel Jarre, he's got a bit of everything in there.

True, I didn't think of him because he tends to bury things and reverb, chorus, and phase shifting, I thought someone would complain it wasn't pure noise, pure square waves, that's why Kraftwerk came to mind, being old and simple electronics...

 

[jensgk]

MQA is not lossless hence everything you said afterwards doesn't have a meaning.
It should be by now a known fact about MQA.
You missed what you are arguing for.

Actual quotes from MQA:
"Q. Is MQA Lossless?
A. Yes."

"Q. Is MQA lossless?
A. It's better than lossless! "

 

[JohnYang1997]

Actual quote from MQA:
"
Q. Is MQA Lossless?
A. Yes.
"

So what? We should have all accept that this is false.

 

[jensgk]

So what? We have all accept that this is false.

So it is snake oil and broken, given their marketing.

 

[JohnYang1997]

So it is snake oil and broken, given their marketing.

Do we care about how MQA actually performs or do we care about bullshit marketing talks?

 

[Raindog123]

All (and especially those in the DAC design business):

I asked @JohnYang1997 yesterday, now broadening my query...

Do the audio DACs (chips and/or designs) have internal "digital loopback" capability? Where the outbound PCM sample stream is pushed through the FIR filter stage (to its 'final raw form' that controls the DAC) and then can be looped-back to the local baseband interface? Where it can be captured and analyzed?

This test 'trick' is widely used in the RF modem designs, I wonder if there is something similar here?

 

[THW]

So it is snake oil and broken, given their marketing.

i believe i have already pointed this out, but i think its a given that marketing is going to embellish the product in some way. after all everyone wants to sell their product and claim their product is the bestest.

i hardly think MQA is unique in this regard.