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

[PO3c]

Because in ultrasonic band the content of music (just harmonics of increasingly lower magnitude) is NOT 24 bit deep, but as Atkinson's post says, 5 bits, and decreasing even more the higher you move in frequency. The rest is noise below, and empty bits above. And so, they don't need to pack 24 bits below the noise of the first 24Khz band, but instead much, much less (and btw, this is not a mysterious, hidden information; it is explained by MQA in their documents even with annoying detail).

Unless, of course, your signal is white noise of big amplitude or square wave, that instead DOES have very high amplitudes of harmonics in ultrasonic (and also in the high region of the sonic bandwidth). Which, btw, is the reason why Amir tests of DACs increase the band to ultrasonics when measuring square waves. That's why MQA performs bad with that type of signal: it is not intended to process that. One of the many reasons why these tests are flawed.

Btw, I can't figure out what's worse: if Archimago and GoldenEar didn't know that (almost unbelievable); or that they knew it and even then, they prepared a test file knowing the type of failures they would get.

Archimago and GoldenEar set up a test with roots in the most common way we know how to test audio tools for reproducing music. With your way of thinking music has to adopt to MQA's definition of technical standards. That not right is it?
Remember MQA are markeded as high resolution, not as 'some part of the audio spectrum' as we now learn by reading spesifications. So yes, I think it would be correct of them to test as they did, — even if they set up MQA to fail simply by the fact how MQA are marketed.

I do make music using modular synthesizers. Colored noise and square waves are musical tools of the trade. Should I comply with a new audio standard trying to establish new and restricted quality for this artform? I think not, — most people would dismiss MQA and rather move on with their creation using tools that better suit the purpose.

With regards to JA explanation. How would one apply dither to the two half? Would one wait to do it in the decoder after the unfold? But then the non-unfold will suffer, so the encoder will have to do this layer after truncating. But then the decoder truncate this layer again...

On a sidenote. By Qobuz own definition any audio file with 24 bit depth are to be considered Hi-Res content. This from their leagal pages.
Then we find 24b MQA encoded music on their platform wrongly labelled Hi-Res without any warning that special licensed hardware or software are needed for 'hi-res' playback.
What we are seeing are MQA sneaking its way into our audio chain like a trojan horse with streaming providers silently comply with the new standard even when it actually are ruining their whole business scheme. If anything, — what MQA has taught us is that 24b audio are now a dated hi-res scam.

 

[sandymc]

Then we find 24b MQA encoded music on their platform wrongly labelled Hi-Res without any warning that special licensed hardware or software are needed for 'hi-res' playback.
What we are seeing are MQA sneaking its way into our audio chain like a trojan horse with streaming providers silently comply with the new standard even when it actually are ruining their whole business scheme. If anything, — what MQA has taught us is that 24b audio are now a dated hi-res scam.

Yes. Exactly this. Previously I thought that MQA was harmless. It's not.

 

[TurbulentCalm]

If you're not seeing any spectral content above 22 kHz, then the MQA hasn't been decoded. Could you post a screenshot of the spectrum you're seeing?

Sure, if I can. Remember this is only NeutronMP’s inbuilt display and is only momentary. I’m not sure if there is anyway I could do this better on my iPhone unless I break out my ancient MacBook.

 

[awdeeoh]

There is now a Hifi Plus in addition to Hifi.

 

[JSmith]

There is now a Hifi Plus in addition to Hifi.

JSmith

 

[RichB]

Without mentioning lossing EQ capability. Since "bit perfect" connection to the DAC is expected! I really do not understand how @amirm is ok with this, him being a proponent of EQ

As I understand it, Roon got dispensation.
No problem, so long as MQA approves of it.

- Rich

 

[RichB]

*** Yikes! This post is being interrupted because I just noticed I had NeutronMP set with 4x oversampling and I expect this is why the sampling rates where so high. I’ll double check everything and update as required. Your regular post will resume shortly with no thanks to this user’s stupidity though. ***

@@@ We are now returning your to your regular post. I have found that only the MQA-CD file’s sampling rate was effected. I’ve updated the screen shot but to summarise: the unfolded MQA-CD is now showing as a 24bit/44.1kHz. That means that other than changing the bit depth from 16 to 24 no other changes have been made. The more I see of MQA the more it appears to be nothing more than a fancy shell game. @@@

Ok I’m sure everyone knows this already but just in case some don’t.

Did you know NeutronMP appears to have reverse engineered the full unfolding of MQA FLAC and MQA CD files?

Using one of the downloadable 2L tracks: Hoff Innocence

View attachment 126537

Exhibit A (MQA-CD):

View attachment 126538

What NeutronMP sees the raw file as:

View attachment 126539

What the (Non MQA) DAC sees (Image updated with correct sample rate — the idiot responsible for the initial debacle has been shot):

View attachment 126562

Exhibit B (MQA Original Resolution):

View attachment 126541

What NeutronMP sees the file as:

View attachment 126546

What the DAC sees:

View attachment 126545

Sorry to bore you all with all this but…

The file is not 352..8 kHz / 24 bit audio, any more than MP3 is 44.1/16, except MP3 is honest about it. It does not proclaim to the 88.2 kHz for example.

- Rich

 

[voodooless]

Here is some more MQA disection: https://www.hifilink.fr/wp-content/uploads/2020/04/MQA-Technical-Analysis-Hypotheses-Paper.pdf

 

[PO3c]

As I understand it, Roon got dispensation.
No problem, so long as MQA approves of it.

- Rich

My understanding 48kHz are the limit MQA allow for digital out. But even if ROON got an dispensation to manipulate higher SR it has to decode the file, apply DSP then encode the file with MQA compression again. Corrected: - ROON can render MQA without the need for compression.

Here is the thing. Even if the connector label in ROON says so there is no way ROON can reconstruct any MQA sound data from the original file other than tags after the DSP event.

Firstly ROON had its way with DSP on the file then it's chewed yet again through the MQA encoder engine to be able to sendt 88.2/96 file ahead in digital form. Then the master authetication should not lid on our DAC should it?

 

[mash]

I switched to Roon and Tidal last year and started out with one of my endpoints as MQA enabled (PC streaming to a D90), Since then, I've backed off of MQA (more on principal than anything else). and I've moved to non-MQA endpoints.

I have to say.....one of the more negative points on MQA (from a marketing POV) is the complexity. There's just no way to really explain it to a casual audio listener other than, " connect your box and hope to see the blue light". From my POV, this is another classic business blunder where proponents are placing bottom line "revenue loss" concerns ahead of top line revenue expansion opportunity.......that approach always loses out in the long run, while it typically does damage in the short run.

That said, here is the signal path according to Roon where EQ is used in conjunction with an MQA DAC. In the path you can see both unfolds and the insertion of the DSP EQ bands into the middle of the process.

 

[mansr]

My understanding 48kHz are the limit MQA allow for digital out. But even if ROON got an dispensation to manipulate higher SR it has to decode the file, apply DSP then encode the file with MQA compression again.

That is incorrect. Roon runs the file through the MQA decoder, producing a 96 kHz signal. This is then processed and marked for MQA "rendering" which is trivial. That is what goes out to the DAC.

 

[pjug]

Has anyone seen any Hi-Res tracks that actually contain ultrasonic frequencies that have also been encoded as MQA?
[/USER].

This plot of dxd and mqa versions of a 2L recording (blue and red in thetop chart in attached image)shows that decoded MQA is faithful to the original to 48KHz. I found this in the @mansr post here: https://audiophilestyle.com/forums/topic/30381-mqa-is-vaporware/page/185/

However, I can't produce the same thing using the MQA decoder in Audirvana. My results don't track to 48KHz (bottom chart). Maybe I flubbed a setting. @Archimago showed that decoders are not necessarily the same. What does that say about MQA if decoding if the first unfold decoding isn't consistent? http://archimago.blogspot.com/2017/02/comparison-hardware-decoded-mqa-using.html

 

[RichB]

Archimago and GoldenEar set up a test with roots in the most common way we know how to test audio tools for reproducing music. With your way of thinking music has to adopt to MQA's definition of technical standards. That not right is it?
Remember MQA are markeded as high resolution, not as 'some part of the audio spectrum' as we now learn by reading spesifications. So yes, I think it would be correct of them to test as they did, — even if they set up MQA to fail simply by the fact how MQA are marketed.

I do make music using modular synthesizers. Colored noise and square waves are musical tools of the trade. Should I comply with a new audio standard trying to establish new and restricted quality for this artform? I think not, — most people would dismiss MQA and rather move on with their creation using tools that better suit the purpose.

With regards to JA explanation. How would one apply dither to the two half? Would one wait to do it in the decoder after the unfold? But then the non-unfold will suffer, so the encoder will have to do this layer after truncating. But then the decoder truncate this layer again...

On a sidenote. By Qobuz own definition any audio file with 24 bit depth are to be considered Hi-Res content. This from their leagal pages.
Then we find 24b MQA encoded music on their platform wrongly labelled Hi-Res without any warning that special licensed hardware or software are needed for 'hi-res' playback.
What we are seeing are MQA sneaking its way into our audio chain like a trojan horse with streaming providers silently comply with the new standard even when it actually are ruining their whole business scheme. If anything, — what MQA has taught us is that 24b audio are now a dated hi-res scam.

Hi-Res audio label is already loosey goosey. Dr. Mark Waldrep has pointed out that masters that do not contain ultrasonic information can be labeld HD-Res Audio as defined by Sony requires the following:

All What you need to know about Hi-Res Audio logo? (samma3a.com)

Japan Audio Society | Hi-Res Audio Logo (jas-audio.or.jp)

1. "Hi-Res Audio" logo applicable products JAS defines is to fulfill the following specification on the recording, reproduction and signal transition process<Analogue process>
   • (1) Microphone response performance: 40 kHz or above during recording
   • (2) Amplification performance: 40 kHz or above
   • (3) Speaker and headphone performance: 40 kHz or above
2. <Digital process>
   • (1) Recording format: Capability of recording using the 96kHz/24bit format or above
   • (2) I/O (Interface): Input/output interface with the performance of 96kHz/24bit or above
   • (3) Decoding: File playability of 96kHz/24bit or above (FLAC and WAV both required)
     (In case of self-recording equipment, FLAC or WAV file is required as minimum condition)
   • (4) Digital Signal Processing: DSP processing of 96kHz/24bit or above
   • (5) D/A conversion: Digital to analog conversion processing of 96 kHz/24 bits or above

You do not see the Hi-Res logo on Tidal.
MQA is not Hi-Res audio, it does not provide 24 bit audio and it is not sampled at the rates shown by the MQA DACs.

- Rich

 

[mansr]

What does that say about MQA if decoding if the first unfold decoding isn't consistent?

At one point, I compared the Bluesound and Tidal (Windows app) decoders, and the output was bit for bit identical. Different versions of the decoder might differ ever so slightly due to rounding errors, but they should be for all practical purposes identical.

 

[pjug]

At one point, I compared the Bluesound and Tidal (Windows app) decoders, and the output was bit for bit identical. Different versions of the decoder might differ ever so slightly due to rounding errors, but they should be for all practical purposes identical.

Thanks. Did you make the chart in the AS post yourself? Would that have been with the Bluesound decoder? Do you have Audirvana to see if you can the MQA first unfold to track to 48KHz?

 

[RichB]

I suppose one of the trouble aspects is the supporters of MQA do not acknowledge the danger posed by monopolization.
They ignore that loss incurred by those who do not upgrade to pay the MQA royalties.

I don't understand motivation by some of the defenses.
Perhaps it is not fair to suspect elitist cronyism or saving face as a motives, but there is not much left.

If defenders acknowledged of lousiness (oops , lossiness), lack of openness, PEQ/REQ restrictions, the concerns both technical and business perspective, then it would garner greater respect.

- Rich

 

[mieswall]

Other than via PowerPoint Engineering, seen to much of that from MQA, or technical statements expressed only in words in this forum, can you please provide links to any technical analysis that demonstrates and verify your claims that you know of?

Thanks.

I would start by this convention paper (free download from AES):
https://www.aes.org/e-lib/browse.cfm?elib=17501
Perhaps a bit difficult to understand it in full, but it gives a good general approach.
There are some 90 references in this paper, some of them could be useful for further reading if you are doubtful about some claims; specially neuroscience research about human perception of time domain.


Then (borrowing some info of a previous comment):
About time domain:
https://www.soundonsound.com/techniques/mqa-time-domain-accuracy-digital-audio-quality
https://patents.google.com/patent/US10867614B2/
https://patents.google.com/patent/US20200395918A1

About file size reduction:
https://patents.google.com/patent/US9548055B2
https://bobtalks.co.uk/blog/mqaplayback/origami-and-the-last-mile/
https://patents.google.com/patent/US20160005411A1
https://www.meridian-audio.info/public/mlp_jap_new[2521].pdf


General articles about MQA:
- Search for a series of articles of 2017, 2018 about MQA in Stereophile by Jim Austin, each one dealing with specific subjects of the technology.
- Nice summary of MQA byRobert Hartley in The Absolute Sound https://www.theabsolutesound.com/articles/beyond-high-resolution
- A very interesting assessment of MQA by Robert Hartley. This I could find it complete only at MQA'ssite, not in TAS:
https://mqa-production.s3.amazonaws.com/default/0001/01/6a10f3ba2385770ac3658df2cadc537ffcd09cd3.pdf
- look here for a John Atkinson's post with an interesting link about an article of him in Stereophile.
- https://www.stereophile.com/content/mqa-questions-and-answers (some 20 pages, all of them worth a open-minded read).

And, I know it is a hard thing to ask, but most of the articles by Stuart here are required reading:
https://bobtalks.co.uk/blog/science-mqa/mqa-playback/#

 

[levimax]

I suppose one of the trouble aspects is the supporters of MQA do not acknowledge the danger posed by monopolization.

Mr. Dorsey, who is one of the richest people in the world, did not purchase Tidal unaware of MQA. He is not going to be content with 1% market share. His other businesses, Twitter and Square, have achieved near monopolies in their respective fields. Whether or not he can succeed in monopolizing the music delivery business remains to be seen but it is a good bet he is going to take a shot at it. I find it interesting that MQA, which has been a smoldering issue for years, has suddenly become hot topic after Mr. Dorsey purchased Tidal. While on these boards MQA is not popular I see a lot of subjective praise about "MQA sound quality" on the other sites. I doubt this in a coincidence.

 

[mieswall]

This plot of dxd and mqa versions of a 2L recording (blue and red in thetop chart in attached image)shows that decoded MQA is faithful to the original to 48KHz. I found this in the @mansr post here: https://audiophilestyle.com/forums/topic/30381-mqa-is-vaporware/page/185/

However, I can't produce the same thing using the MQA decoder in Audirvana. My results don't track to 48KHz (bottom chart). Maybe I flubbed a setting. @Archimago showed that decoders are not necessarily the same. What does that say about MQA if decoding if the first unfold decoding isn't consistent? http://archimago.blogspot.com/2017/02/comparison-hardware-decoded-mqa-using.html

View attachment 126623

And how do you get with Audirvana the additional unfolds by hardware? Perhaps what you are seeing here is just the first unfold, the only one possible to be done by software. 2L files unfold to 352K sampling rate. If you are not decoding by hardware, you are not getting a full file reconstruction intended by MQA. In my experience, decoding by hardware with a MQA DAC makes a BIG sonic difference.

 

[pjug]

And how do you get with Audirvana the additional unfolds by hardware? Perhaps what you are seeing here is just the first unfold, the only one possible to be done by software. 2L files unfold to 352K sampling rate. If you are not decoding by hardware, you are not getting a full file reconstruction intended by MQA. In my experience, decoding by hardware with a MQA DAC makes a BIG sonic difference.

I'm only wanting to get the first unfold, trying to reproduce the plot posted by @mansr where first unfold tracked the hires to 48KHz.