John Atkinson: Yes, MQA IS Elegant...

[tmtomh]

I find it noteworthy that @John_Atkinson's article frames itself as a response to the criticism that he's called MQA "elegant." Given that "elegance" isn't the description for which Atkinson has gotten the bulk of the criticism directed at him in what he calls the "fracas" over MQA, it appears he's cherry-picking the criticism he wants to defend himself against, which has the ancillary effect of absenting all other criticisms of his pro-MQA statements from the article.

 

That said, I think his article is a very well-written, informative recitation of the case for minimizing pre-ringing (aka prioritizing time-domain accuracy) in digital filtering. Anyone who's not already well-versed in this issue (either by professional training or from doing a lot of reading) would, I think, benefit from it.

 

There does seem to be one major logical (or perhaps evidentiary) gap in the piece, though: He cites a listening test that failed to turn up audible evidence that pre-ringing causes problems (except in an intentionally worst-case filter where all the ringing is pre-ringing). Then almost immediately he nevertheless speculates that pre-ringing nevertheless could be an issue (by wondering if higher sample rates are pleasing to folks because they move ringing out of the audible range) - and then quickly moves on from there to strongly imply that pre-ringing is indeed a problem, despite the lack of evidence in the listening test mentioned just a couple of paragraphs above.

 

Folks can of course read the piece for themselves, so they don't have to blindly accept my characterization of it - the portion I'm referring to is on Page 2 of the web article linked to in the first post of this thread.

[tmtomh]

TAS in particular, and Stereophile to a degree, always (or at least for a long time) have been defined subjective, impressionistic listening tests. This means that they will tend not to say that new or interesting-sounding technologies, upgrades, or innovations don't make any sonic difference. The reason is simple: If you feel that MQA doesn't improve the sound, is that because MQA isn't all it's cracked up to be, or because your hearing and listening skills (or your reference rig) are not developed and discerning enough to detect the differences? If someone else, especially at a peer/rival publication, claims to hear a difference after you've just published a piece saying you couldn't hear a difference, you risk demoting your reputation.

 

I am not saying that reviewers consciously think in these terms - in fact, I would guess they don't. Instead, I would guess they feel that they go into this stuff with an open mind, and consistently are "struck" or "surprised" to find that they do in fact hear differences.

 

IMHO - and I stress this is just my opinion - they underestimate the power of confirmation bias (which doesn't have to manifest itself as a conscious "I need to hear a difference here to burnish my golden-ears rep" feeling), and they overestimate the reliability of human auditory memory when it comes to fine details over extended listening periods in a non-A/B (or non-A/B/X) situation.

 

My point here is that their cheerleading for MQA is not necessarily that much of an outlier from their usual approach to tech claims made by manufacturers and equipment designers. The overall attitude required to compare power cords, and to compare 96kHz vs 192kHz sample rates, is very similar.

[tmtomh]

 

2 hours ago, John_Atkinson said:

 

It saddens me that so much of this thread has degenerated into personal attacks, from people who have not read what I have written or, if they have done so, have not understood it, as in the case of the post which I am responding. Too many posters have a greater confidence in their opinions than they do actual knowledge, driven by what appears to be antagonism toward those of us who earn our livings by what we publish and write.

 

[...]

 

The experimental evidence I presented is incontrovertible. That unless the user of an A/D converter is prepared to accept the possibility of some aliased image energy in order to use an antialiasing filter that preserves the time-domain behavior of the original analog signal, the resultant digital data will have sinc-function content at the Nyquist frequency accompanying every musical transient.

 

[...]

 

If a specific type of slow-rolloff antialiasing filter in the A/D converter is combined with a reconstruction filter in the D/A converter that behaves in a similar manner, you will have an analog-digital-data storage/transmission-digital-analog chain that will have an impulse response and top-octave rolloff equivalent to that of a small distance of air. As I wrote in my article, this was the stated goal of both Bob Stuart and the late Charley Hansen; I see no disrespect to the latter in pointing that out, especially as I instanced Charley's antipathy to MQA in the article.

 

[...]

 

Is the compromise in the frequency domain associated with using time-domain-perfect converters acceptable with real musical signals? Or is the possibility of image energy being aliased into the audioband too great? I have heard arguments on both sides of this question. I suspect the answer is that it depends on the type of music.

 

Does ringing at 22.05kHz even matter when it comes to sound quality? Again, I have heard arguments on both sides of this question, from people I respect.

 

@John_Atkinson, thank you for your thorough comment here (which I upvoted). I don't agree with all of it, but I certainly agree that some folks here are resorting to personal attacks and general snark, and in the process avoiding substantive debate.

 

I hasten to add, though, that while the debate you and @mansr are in is antagonistic in tone, I believe it's false to claim that he's making a personal attack. He's making a specific claim - which he's been consistent about in this and other MQA threads, and which he's been very specific about in multiple prior comments - that some of what you are claiming does not comport with his understanding of the relevant mathematical principles.

 

For my part, I don't have the level of technical expertise to weigh in on the specific issues up for debate. I will only say that I do think there's a specific issue with the ideas as you express them in your Stereophile article (which article I've praised earlier in this thread).

 

The issue is that you see to be inconsistent in how you write about the audibility of pre-ringing. In the piece, you cite a listening test that failed to turn up evidence that it's audible (except for the intentionally-poorly designed filter). But then you speculate, as you do in the comment above, that it might possibly cause problems - and then almost immediately thereafter in the article you strongly imply that it is indeed a problem.

 

It seems to me that the case for MQA - whether one calls it "a whole new world" or "elegant" or simply "good" or "important" - depends upon pre-ringing being a major issue. And your discussion seems to zig towards asserting this when you are discussing MQA and its beneficial/intriguing/elegant design, and then zag towards "who knows, maybe it's not an issue at all" when you are explicitly critiqued for your apparent editorial stance on MQA.

 

This is not a mathematical critique, but it's not a personal attack either. I'm making a rhetorical observation and objection, that to my eyes is about the way you have framed coverage of MQA. I'd welcome your response and thoughts. Thank you.

 

 

On 8/22/2018 at 5:35 PM, tmtomh said:

[...] I think his article is a very well-written, informative recitation of the case for minimizing pre-ringing (aka prioritizing time-domain accuracy) in digital filtering. Anyone who's not already well-versed in this issue (either by professional training or from doing a lot of reading) would, I think, benefit from it.

 

There does seem to be one major logical (or perhaps evidentiary) gap in the piece, though: He cites a listening test that failed to turn up audible evidence that pre-ringing causes problems (except in an intentionally worst-case filter where all the ringing is pre-ringing). Then almost immediately he nevertheless speculates that pre-ringing nevertheless could be an issue (by wondering if higher sample rates are pleasing to folks because they move ringing out of the audible range) - and then quickly moves on from there to strongly imply that pre-ringing is indeed a problem, despite the lack of evidence in the listening test mentioned just a couple of paragraphs above.