MQA is Vaporware

[mansr]

4 hours ago, Jud said:

I don't know enough to know how much or even if MQA's filters screw up the time domain, but this from the linked review caught my eye:

 

"The math for this MQA 'trick' can get very complex and I won't try to explain it, but B-spline interpolation is very modern and the Shannon-Nyquist theorem was first published in 1928, a long time before Sony and Phillips developed the Compact Disc."

 

Yeah, and the Pythagorean Theorem is 2500 years old, so that’s probably no longer true either.  (OK, non-Euclidean geometry, but you get the point.) Jeez, attempts at simplified mathematical or scientific explanations that are just plain wrong are irritating.

 

It's not possible in the real world to make the perfect filter you'd need for Shannon-Nyquist to work perfectly, so yeah, better and worse digital filters are possible, or filters that lean one way or the other in terms of having most of their imperfections in what we think of as the time or frequency domains.  But B-spline interpolation doesn't exempt you from the rules of math. 

B-splines are just another method for approximating ideal reconstruction. Compared to the usual windowed sinc filter, the resulting error characteristics are a bit different. For audio purposes, these differences are utterly irrelevant. The "post-Shannon" developments are aimed at recovering specific aspects of the signal with less computational effort than a full sinc interpolation would require. Again, this is not necessary for audio since the complexity involved is trivial to begin with.

[FredericV]

2 hours ago, Shadders said:

"However, this section will argue that a phase response somewhere between linear- and minimum-phase may be even better in some cases"

 

Regards,

Shadders.

 

We did test intermediate phase vs linear & minimum phase on several cost is not object systems, and intermediate phase outperforms all the others to our ears.

http://archimago.blogspot.com/2018/01/musings-more-fun-with-digital-filters.html

 

[Jud]

1 hour ago, FredericV said:

 

We did test intermediate phase vs linear & minimum phase on several cost is not object systems, and intermediate phase outperforms all the others to our ears.

http://archimago.blogspot.com/2018/01/musings-more-fun-with-digital-filters.html

 

 

Yes, but it doesn't mean intermediate phase, rather than minimum or linear phase, is best for any given system.  And it also doesn't mean it will be best for your individual hearing and preferences.

 

By the way, intermediate phase filtering screws up what you spoke of as time domain response (though I've never read anyone else describing these as time domain distortions) too, just a bit less so than minimum phase filtering.

 

It's just no good being doctrinaire about this.  Different types of filters suit different situations.

[Jud]

2 hours ago, mansr said:

IIR or FIR has nothing to do with the phase response. All the usual phase responses can be realised as either type.

 

Infinite *impulse* response and finite *impulse* response, so as you say, not related to phase. 

 

https://en.m.wikipedia.org/wiki/Finite_impulse_response

 

https://en.m.wikipedia.org/wiki/Infinite_impulse_response

[mansr]

2 minutes ago, Jud said:

Infinite *impulse* response and finite *impulse* response, so as you say, not related to phase. 

 

https://en.m.wikipedia.org/wiki/Finite_impulse_response

 

https://en.m.wikipedia.org/wiki/Infinite_impulse_response

An IIR filter uses feedback from the output which allows a finite-length filter to produce an infinite-length impulse response. For instance, a step response is trivially realised. A FIR filter used only values from the input, so the impulse response of a finite-length filter will eventually return to zero.

[Jud]

4 hours ago, Rt66indierock said:

 

A while back I had said I wanted control of filters.  Julius O Smith III (the Stanford source) has said you need minimum phase, intermediate phase and linear phase filters. I agree with his assessment so far. 

 

Player software like Audirvana Plus or HQPlayer, or audio software like iZotope or SoX, offers such choice/control.

 

4 hours ago, Rt66indierock said:

As I recall you have Vandersteen speakers. I listened to a pair briefly at T.H.E. Show.

 

That's right.  I've owned Vandersteens for 30 years, so when I encounter them at a show, it's "Ah, this sounds right."  Of course someone with a different history could easily have a different reaction.

[botrytis]

15 hours ago, Jud said:

 

Player software like Audirvana Plus or HQPlayer, or audio software like iZotope or SoX, offers such choice/control.

 

 

That's right.  I've owned Vandersteens for 30 years, so when I encounter them at a show, it's "Ah, this sounds right."  Of course someone with a different history could easily have a different reaction.

Like me. i couldn't get rid of them fast enough.

[Jud]

4 hours ago, botrytis said:

Like me. i couldn't get rid of them fast enough.

 

What speakers do you like?

[botrytis]

2 hours ago, Jud said:

 

What speakers do you like?

Honestly, the best system I heard at AXPONA was using Dynaudio C2 Platinum speakers which I really like. I also like the new Bowers and Wilkins  800 D3 series, especially the 804 D3 which sounded stellar.

[firedog]

On 6/14/2018 at 5:57 PM, FredericV said:

So this article is more BS. Steven Stone just keeps pushing articles like this on audiophile Facebook groups:
 

https://audiophilereview.com/cd-dac-digital/mqa---its-about-time-not-frequency.html

Read the article and it is bunk, I made a few comments, and the author can’t even admit that MQA is lossy. His only response was hand waving and insults. 

[Brinkman Ship]

2 hours ago, firedog said:

Read the article and it is bunk, I made a few comments, and the author can’t even admit that MQA is lossy. His only response was hand waving and insults. 

The author is a joke. Not even worth 2 seconds of consideration. In fact, this "publication: is a total and utter joke.

[MikeyFresh]

4 hours ago, firedog said:

His only response was hand waving and insults.

 

His response was both pathetic, and also seemingly not his own, dare I say the "article" was just a paraphrasing of some Bob Stuart penned PR copy and the comment responses directly coached by someone within MQA? 

 

I am surprised that at this late juncture, thats all they have in the form of supporting argument. It amounts to nil, and that particular author goes up in flames right there and then with his non-response to your well stated comments.

[fung0]

On 6/7/2018 at 11:39 PM, Brinkman Ship said:

For the record, i do not believe Stereophile writers are paid to mention products ("product placements").

 

I don't think John Atkinson would allow it. TAS..it is possible..but only with certain writers.  But still unlikely. I don't think audio manufacturers would really need to do this. Providing the products for long term loan is sufficient enough incentive to get them mentioned.

 

I recently received an email in response to an article of mine that was published online. The email was from a commercial service that monetizes links. The idea was that if I signed up with them, I could make money every time someone followed a product link included in one of my articles. Clearly, there are companies making a business out of 'rewarding' journalists in this way. So presumably there are takers. It doesn't mean their reviews are slanted, necessarily... but it's not exactly the kind of incentive that would encourage incisive journalism.

 

Regarding long-term loans: yes, they're always a risk. Reviewing tech gear doesn't pay remotely enough to cover the expense of keeping one's own equipment close to the leading edge. So it's extremely tempting to accept 'loaners' or discounted purchases. In my experience, most such insider deals do not have specific strings attached. (Maybe it's just that I've never expressed enough openness to the idea. Or that I've never reviewed anything in the five- to six-figure price bracket.)

 

Also, don't forget that there's both a carrot and stick. Companies definitely tend to become less responsive to journalists who are repeatedly critical of their products or policies. Being shunned by a major industry player can be disastrous for a working tech journalist.

[asdf1000]

On 6/15/2018 at 3:49 AM, Brinkman Ship said:

I am going to be doing a direct comparison between a CD Rip of:

 

 

 

and the 24/44.1 MQA version

 

Please not again. Remember that time you started a thread? And the Brinkman Ship suffered the same fate of the Titanic? ?

 

 

[Brinkman Ship]

6 hours ago, Em2016 said:

 

Please not again. Remember that time you started a thread? And the Brinkman Ship suffered the same fate of the Titanic? ?

 

 

Oh it's on...CDs ripped....

[Brinkman Ship]

8 hours ago, fung0 said:

 

I recently received an email in response to an article of mine that was published online. The email was from a commercial service that monetizes links. The idea was that if I signed up with them, I could make money every time someone followed a product link included in one of my articles. Clearly, there are companies making a business out of 'rewarding' journalists in this way. So presumably there are takers. It doesn't mean their reviews are slanted, necessarily... but it's not exactly the kind of incentive that would encourage incisive journalism.

 

Regarding long-term loans: yes, they're always a risk. Reviewing tech gear doesn't pay remotely enough to cover the expense of keeping one's own equipment close to the leading edge. So it's extremely tempting to accept 'loaners' or discounted purchases. In my experience, most such insider deals do not have specific strings attached. (Maybe it's just that I've never expressed enough openness to the idea. Or that I've never reviewed anything in the five- to six-figure price bracket.)

 

Also, don't forget that there's both a carrot and stick. Companies definitely tend to become less responsive to journalists who are repeatedly critical of their products or policies. Being shunned by a major industry player can be disastrous for a working tech journalist.

There is no doubt what you detail is common...

 

I will note however that I can't remember seeing product links in any high end audio reviews. Links back to previous reviews,

yes, but not click bait in general.

[firedog]

19 hours ago, MikeyFresh said:

 

His response was both pathetic, and also seemingly not his own, dare I say the "article" was just a paraphrasing of some Bob Stuart penned PR copy and the comment responses directly coached by someone within MQA? 

 

I am surprised that at this late juncture, thats all they have in the form of supporting argument. It amounts to nil, and that particular author goes up in flames right there and then with his non-response to your well stated comments.

Yeah, he keeps insisting that the MQA process is lossless:

 

Quote

. As I have said in other posts, MQA is lossy until you fully decode as opposed to unfold it where B-spline interpolation is used in conjunction with a compressed reference signal to reconstruct the original analog waveform. I’m sorry that this is beyond your own imagination or intellectual capacity to understand, but those as are the facts.

 

[mansr]

9 minutes ago, firedog said:

Yeah, he keeps insisting that the MQA process is lossless:

 

Quote

As I have said in other posts, MQA is lossy until you fully decode as opposed to unfold it where B-spline interpolation is used in conjunction with a compressed reference signal to reconstruct the original analog waveform. I’m sorry that this is beyond your own imagination or intellectual capacity to understand, but those as are the facts.

That's both nonsensical and insulting. Engaging with such people is a waste of time.

[fung0]

13 hours ago, Brinkman Ship said:

There is no doubt what you detail is common...

 

I will note however that I can't remember seeing product links in any high end audio reviews. Links back to previous reviews, yes, but not click bait in general.

 

It's true, I haven't run into any links that obviously use this particular monetization scheme. On the other hand, most links these days are coded in ways that would make it difficult to know if something like this were going on.

 

The larger point is that there is a very real, long-standing movement to monetize journalism. One of the reasons I find the MQA situation so fascinating is that it highlights as never before the rot that's long been growing at the heart of tech journalism in particular.

 

When I started out in print, about half the income of the magazine I worked for came from the readers - and much of that via annual subscriptions, which provided a comfortable revenue base. Today, few web publications are able to generate much direct consumer revenue. Readers are monetized via advertising, of one form or another. As is often pointed out in the case of Facebook, this means that consumers are no longer the customers - they are now the product that is sold to the actual corporate backers.

 

Even within this broken system, individual journalists are rarely evil people - and overt corruption is probably much rarer than any of us might estimate. But the incessant pressures of the business make it harder and harder to take a truly unbiased stance. Digging deep is not rewarded, and expertise is valued less than a glib ability to sound knowledgeable while saying nothing of substance.

 

MQA seems like the culmination of all that. It's not the only recent example of a marketing campaign masquerading as a technological breakthrough - but it is surely the leader in its field. It's got something for every audio journalist - excepting maybe a vanishingly few who still dare to ask tough questions, or challenge extravagant claims. (It's not accidental that the main challenges to the MQA narrative have come from people who do not earn their living this way.)

[FredericV]

9 hours ago, firedog said:

Yeah, he keeps insisting that the MQA process is lossless:

 

 

Andy does not understand MQA at all.
 

Quote

 

Andy Schaub  firedog55 • 9 hours ago

I’m afraid that you’re incorrect. You’re confusing unfolding to 24/192 or higher which is lossy without full decoding, something that has to be done in the DAC itself using B-spline interpolation and the compressed reference signal, masked by dither, to restore the full analog waveform in so much as any digital system can (although both signals are 24/192, the unfokded signal is lossy but not the fully “decoded” one).

 

 

 

You cannot unfold to 24/192 without full decoding. If some MQA decoder delivers 24/192, by design it must be doing full decoding:

- first it does the first unfold, to something like 17/96 packed in 24/96
- then it upsamples this in the renderer to 24/192, which is the second unfold

The only exception could be rare source material in 24/384, where your MQA dac is limited to 24/192 only, so it does not upsample to the original resolution, but to the maximum multiple of 48 Khz the DAC can do.

And MQA is ALWAYS lossy.

And OMG he can't even repeat the MQA marketing:

 

Quote

Andy Schaub  firedog55 • 7 hours ago

... What you fail to understand and just won’t try to comprehend is that traditional compression occurs in the time domain whereas MQA compression works in the frequency domain. So a fully end-to-end MQA chain is lossless however you have misinterpreted what you read (and which you still won’t cite).

MQA obsesses over the time domain, therefore giving up accuracy in the frequency domain:



 

[firedog]

4 minutes ago, FredericV said:

Andy does not understand MQA at all.
 

 

You cannot unfold to 24/192 without full decoding. If some MQA decoder delivers 24/192, by design it must be doing full decoding:

- first it does the first unfold, to something like 17/96 packed in 24/96
- then it upsamples this in the renderer to 24/192, which is the second unfold

The only exception could be rare source material in 24/384, where your MQA dac is limited to 24/192 only, so it does not upsample to the original resolution, but to the maximum multiple of 48 Khz the DAC can do.

And MQA is ALWAYS lossy.
 

Yes, but he has repeatedly insisted that " B-spline interpolation is used in conjunction with a compressed reference signal to reconstruct the original analog waveform" and that I don't know what I'm talking about. I asked him what an MQA encoder does with audio signals above 48khz. We know that it doesn' t  encode them, something which seems to escape him. 

 

I'd guess you could get a non lossy MQA file if you had a very bandwidth limited original source.

[FredericV]

1 minute ago, firedog said:

Yes, but he has repeatedly insisted that " B-spline interpolation is used in conjunction with a compressed reference signal to reconstruct the original analog waveform" and that I don't know what I'm talking about. I asked him what an MQA encoder does with audio signals above 48khz. We know that it doesn' t  encode them, something which seems to escape him. 

 

I'd guess you could get a non lossy MQA file if you had a very bandwidth limited original source.

How can he know the original analog waveform is intact after applying the claimed MQA end-to-end process, when the tools to assess these claims are based on regular PCM which MQA tries to fight. How can he prove that? He most likely can't.

 

[rickca]

He seems to be saying that full decoding is a different process than core decoding + rendering.  I've never heard that before.

[FredericV]

1 hour ago, rickca said:

He seems to be saying that full decoding is a different process than core decoding + rendering.  I've never heard that before.

Maybe we should apply MQA to Andy's comments to save bandwidth, which he keeps repeating when he looses the discussion, like this one about Archimago's article:
 

Quote

"That article is essentially an op ed based on a collection a random and orthogonal points intended to prove a point rhetorically rather than deduce a conclusion logically and relies heavily on quotes from Bob or a lack of comment from Bob to "prove its point". It has no merit whatsoever and makes your comment above ("Bring facts and contrary analysis, not quotes from Bob Stuart") inherently self-contradictory."

Bob has not debunked this article, but instead now we have reviewers attacking the article. How convenient.

[firedog]

2 hours ago, rickca said:

He seems to be saying that full decoding is a different process than core decoding + rendering.  I've never heard that before.

He is an interesting case. Acc'd to him, he has tech bona fides and has thoroughly studied the MQA patents and other technical material. Yet he keeps insisting that it isn't lossy, when even BS and MQA admit that it is. He also doesn't get that it's max resolution is about 17bits 96k, and the rest is upsampling. 

I tried to get through to him, didn't work. Apparently he misunderstood how MQA operates, and insists on holding on to his misunderstanding.