Another major look at MQA by another pro.

[mansr]

2 hours ago, plissken said:

I just saw 6 moons / Ebaen and immediately it goes to the ignore department.

He's just publishing an email someone sent him in full.

[mansr]

3 hours ago, NOMBEDES said:

Major point, which is not often expanded upon, is the claim that MQA screws up DSP.  If true, that is a big fly in the ointment.   DSP is the future.  MQA, maybe not so much?

MQA certainly ruins DSP as it normally used today. Before any manipulation, the core decoder must do its job. Here there are two possibilities. The MQA decoder can be part of the DAC, which means the DSP must also be done there. This means we're looking at an AVR-like device incorporating both MQA decoder and DSP stage. Such a device can of course be quite capable, but forget about tinkering with HQPlayer or custom solutions. The other option is software decoding. This produces a 96/24 PCM stream to which DSP can be reasonably applied. Doing so will incapacitate any downstream renderer, but that's actually a good thing. The problem is that software decoders are rare as hen's teeth. Audirvana is the only general-purpose player with MQA support and it runs only on Macs. If you prefer another OS or player, you're out of luck. Others may of course appear with time, but there's no way a free, let alone open source, player will ever be officially licensed. This is not a path we should be taking.

[mansr]

48 minutes ago, soxr said:

At least the strange MQA filters being applied in the renderer phase will not be used by going for the software route and doing just the first unfold.

So this part is basically a dynamic range limited baseband signal (because MQA uses 9 bits for the lossy HF part, encoded as hissing in the baseband signal) + recovered lossy ultrasonics mixed together.

I should still do the test how much diff can be heard between first unfold + sox minimum phase vs undecoded MQA + sox minimum phase.

Thanks to your work we can now dissect all parts of MQA and listen / evaluate each part. The black box is being deblurred part by part

Undecoded MQA has additional noise that isn't part of the encoded upper band. It's just shaped dither from a specific pseudo-random number generator, and the first thing the decoder does is to reverse it. This noise is what gives the characteristic hump in the spectrum from 15 kHz and up.

[mansr]

4 minutes ago, PeterV said:

Nobody over here knows how the MQA encoding and decoding process really works.

I know how the decoder works, and if you'd bothered to read my posts, so would you.

[mansr]

27 minutes ago, PeterV said:

Well, I disagree, the encoding and decoding are much more sophisticated than what is speculated

It's not speculation.

[mansr]

57 minutes ago, PeterV said:

insulting me for being a 'fanboy' being paid by MQA

If you're not paid by MQA, you are the epitome of a fanboy. If you are paid, you're a shill.

[mansr]

2 minutes ago, Fokus said:

Moreover, the renderer only acts on a core-decoded 2x input (ignoring for the moment the pathological case of 1x MQA), where HF signal levels are lowish and thus the output of the 'weird' filters is not that weird at all, if you follow the reasoning behind MQA.

Those "pathological" 1x files make up about half the MQA tracks on Tidal.

[mansr]

14 minutes ago, Fokus said:

Yes. But they are not the reason MQA exists, and nor are they related to the fundamental trickery behind it.

 

It is as if the markering dept saw Craven/Stuart's ideas and then asked "how can we broaden the umbrella to include, and cash in on, 1x as well?"

The reason MQA exists is to put money in Bob Stuart's pocket.

[mansr]

5 minutes ago, Jud said:

When we get beyond the following summary, someone please wake me :

 

- MQA uses sloppy filtering frequency-wise, which allows imaging that creates IMD.  The IMD may be at audible levels, maybe not.

 

- Some people like MQA, some don't.  At least some of the liking may stem from mastering that is different, aside from anything to do with the MQA process itself.

 

(P.S. Where I've heard MQA and hi res of an album that sound like they are from the same master, I subjectively have tended to prefer the hi res slightly or more than slightly, in my system with my settings.)

One more point:

 

- MQA makes DSP EQ and room correction impossible.

[mansr]

3 hours ago, The Computer Audiophile said:

Without a license and working with MQA ltd.

Yeah, good luck with that.

[mansr]

1 minute ago, PeterV said:

The inaudible part is indeed lossy

If it is inaudible, why include it all?

[mansr]

31 minutes ago, Shadders said:

In addition, i thought the blur was the time delay aspect of higher frequencies compared to the lower frequencies- however small - is the issue that MQA is trying to solve ?

If this is what is meant by blurring, MQA can only be making it worse with its minimum phase filters.

[mansr]

17 minutes ago, Fitzcaraldo215 said:

No, the ringing is related to the frequency of the impulse, and the ringing is not at a fixed frequency.

An impulse consists of all frequencies.

 

17 minutes ago, Fitzcaraldo215 said:

Again, the ringing is in the time domain, but it also presents itself in the frequency domain.

Time and frequency are equivalent, dammit.

 

17 minutes ago, Fitzcaraldo215 said:

 MQA have already defined and used the term temporal blur,

Oh, I see it now. Temporal blur means vigorous waving of hands.

[mansr]

3 minutes ago, Shadders said:

HI,

OK - i located the link and it is defined as follows :

"What we mean by temporal blur . . .

There is no standard measure for temporal blur but we believe our use of the term is clear and intuitive.
Read more at,, A causal transmission system has dispersive properties which result from filtering or attenuation. Fine details in the time waveform can be smeared or obscured if the end-to-end impulse response is not sensitive to the signal and to the receiver (human listener)."

Technobabble.

[mansr]

11 minutes ago, Shadders said:

Hi mansr,

Yes - it is technobabble, but in essence they are referring to dispersion - there is no new technical revelation or theory in engineering in MQA. They should just call it dispersion or footnote temporal blur as such.

That blurb you quoted means nothing at all. It is thus impossible to infer what they might actually be talking about. A reasonable guess would be nothing at all. It's all marketing without any substance.

[mansr]

49 minutes ago, Charles Hansen said:

With FIRs it is possible to construct almost any arbitrary filter imaginable - linear phase, minimum-phase, low-pass, high-pass, band-pass, band-reject, et cetera, et cetera. For decades the amount of computational horsepower and memory required made this extremely expensive, but in the last decade of so things have changed radically and that is why you now see companies like Chord making digital FIRs with tens of thousands of taps. (For comparison sake, the digital FIRs built into most DAC chips typically have a couple of dozen taps for cheap parts and as many as a hundred or so taps for expensive parts.)

For additional comparison, the MQA render filters have, for the most part, around 10 taps, a few as many as 20.

[mansr]

4 hours ago, Shadders said:

My interpretation is that temporal blur is dispersion. The article may not be a scientific paper, but the analogies are all referenced to optical dispersion and actually states "Blurring has a direct parallel in the optical world as it relates to the design of lenses, dispersion of light in media, in image processing. In electronics, this is well understood by the designers of oscilloscopes"

There is no such thing as an audio lens. Analogies with optics are tenuous at best.

 

4 hours ago, Shadders said:

My interpretation of this is that they may be obfuscating the issue purposefully.

Without a doubt.

[mansr]

2 hours ago, PeterV said:

NAD encorporates EQ settings together with MQA

If by EQ you mean bass/treble controls implemented as a simple biquad filter, yes.

[mansr]

4 minutes ago, PeterSt said:

Maybe hard to believe because we have learned to think in filtering and taps etc., but a little bit like Arc Prediction (at the sample level)

What is arc prediction? Some kind of spline interpolation?

 

4 minutes ago, PeterSt said:

but at the more macro level : analyse the wave and alter it. No taps but smart averaging per wave chunk (not wave cycle, this can not work).

Anything that takes an input signal and produces an output signal is a filter. All linear filters can be described as a ratio of polynomials in z (poles and zeros) or as a difference equation (taps).

[mansr]

1 hour ago, Jud said:

My possibly faulty concept from what you are saying is that everything working as correctly as possible per the Sampling Theorem requires a properly band-limited signal.

Tautology of the day.

[mansr]

57 minutes ago, Fokus said:

Even more exactly: the 'weird' filters are the anti-imaging filters for the upsampling process.

Except they're not particularly anti.

 

Quote

(And in the original embodiment of MQA these filters were meant to be loaded into the DAC chip, substituting the DAC chip's standard filters. This then would limit MQA to the handful of chips that support this. I can imagine that in some present or future incarnations the upsampling with these filters is actually done in the USB receiver processor preceding the DAC chip, if it is of adequate power.)

The real reason for the hardware tie-in is probably so it can be patented. Under US patent law, a set of numbers can't be patented while a device incorporating those numbers can. Or something like that. This is where @Jud tells me I'm wrong.

[mansr]

6 minutes ago, Jud said:

I do?

You corrected me on some legal matter once.

 

6 minutes ago, Jud said:

I did mention I thought the crypto was essentially for DMCA purposes.

Also a good point.

 

6 minutes ago, Jud said:

Whether mere physical embodiment would help something that can't otherwise be shown to be novel does seem dubious to me, though.

I wasn't thinking of novelty but about eligibility. Unless I'm mistaken, abstract algorithms aren't eligible for patent protection whereas a physical embodiment can be. I thought that was why so many patents describe "a machine doing X" rather than simply "doing X."

[mansr]

7 hours ago, Jud said:

 

You don't need a concrete embodiment (remember there are patents for business methods, for example), but you do need a useful result.  (I'm talking purely about the US, and hasten to add I am not a patent attorney and not giving legal advice.)

 

Edit: https://en.wikipedia.org/wiki/Software_patent#United_States

https://en.wikipedia.org/wiki/Alice_Corp._v._CLS_Bank_International

[mansr]

1 minute ago, Jud said:

And what do you think the case means?

The Wikipedia article has a bit of commentary.

[mansr]

13 minutes ago, Jud said:

And the key, repeated often, is that one can’t patent a mere abstract idea.  Do you think a filter, implemented in hardware or software, is a mere abstract idea, or that it accomplishes a result?

Before Alice it was possible to patent an abstract idea by having it performed by a computer. It's a bit harder now. Of course, patent examiners are as lax as ever, and invalidating an issued patent is a costly legal proceeding.