Article: MQA: A Review of controversies, concerns, and cautions

[miguelito]

53 minutes ago, mansr said:

Interpolation in general means calculating additional values between two given values. Applied to a sequence of samples, this naturally results in an integer multiple.

Also many possible kernels (base functions you use to create the interpolated shape - @mansr: obviously you know). I am not sure how sophisticated this gets in upsampling, but it can get pretty frigging crazy.

[toast]

I hate monopolies!  Thanks you so much for posting this article.

[motberg]

6 hours ago, Doug Schneider said:

. What I do know is that he's not in the industry...

Doug
SoundStage!

 

 

I do not understand your comment above....

 http://archimago.blogspot.com/

I would think this generates revenues?

 

[kumakuma]

3 minutes ago, motberg said:

Archimago owns Madrona Digital and http://archimago.blogspot.com/

 

I think you're confusing Archimago with Amir from Audio Science Review and Madrona Digital.

[MikeyFresh]

4 hours ago, Doug Schneider said:

Sam's real-life alter ego, Tom Gillett

 

And maybe the final word on this, thats the correct spelling of his name to my knowledge, despite various misspellings preceding in this thread (curiously including Kal's).

 

4 hours ago, wdw said:

Tellig was entertaining!

 

That he was, for many years (decades).

 

4 hours ago, wdw said:

To my mind Dudley is in the same realm.

 

Agreed, and sadly now seemingly part of a time gone by. The print magazines used to be something to look forward to, if not always taken with a grain of salt.

[motberg]

11 minutes ago, kumakuma said:

 

 

Yes I was, very sorry and I have edited my post.

I think it would be good if you would please do likewise to remove the false relation

Thanks

[Fokus]

7 hours ago, adamdea said:

The first 14 bit players were called 4 x oversampling I’m fairly sure.

 

That is generally correct, although many years ago I found a very old Philips document in which their 14 bit 4 x approach was named 'upsampling'.

 

To all interested: in the context of consumer audio, oversampling and upsampling always mean (except perhaps in a few freak products) increasing the sample rate of a given digital signal, accompanied with digital-domain anti-imaging/reconstruction/interpolation filtering.

 

The in/out sample rate ratio may be integer or not, in the grand scheme of things that is but a detail.

 

All other 'definitions' such as 'increasing the clock rate' or 'inserting zeroes' are obfuscating and thus useless.

[astromo]

10 hours ago, sullis02 said:

 

Stuart has been flogging, and seeking ways to monetize, hi rez for two decades.   His "Coding Methods for High-Resolution Recording Systems," was presented at the 103rd AES Convention, New York, 1997, and by 2000 became the Meridian white paper "Coding High Quality Audio", which begat his March 2004 JAES paper "Coding for High-Resolution Audio Systems". ...the latter prompting skeptical comments  to JAES by Meyer, Moran, Roy Allison, and David Hardaway a few issues later.  Meyer and Moran's paper was a direct result of Stuart's ( at the time, virtually unsupported by published blind tests) claims.

 

There is lots of history here.  The AES for much of the past two decades was skeptical of the audible benefits of and 'need' for hi rez at the consumer playback end; but the 'political' tide in the organization  on that, seems to have changed in recent years.  There is likely a story there , too.  

 

 

 

Thanks for shedding light on the back story.

 

I raised the Moran and Meyer work a few posts back to contrast with the lack of transparency and empirical validation being offered by the proponents of MQA. It was pointed out that the M&M work was flawed.

 

I don't see that as a problem.

 

For me, that's still a positive. They published their method, theories and results and in time it was critiqued. To their credit, M&M appear to have acknowledged their shortcomings. Other researchers have gained benefit from those mistakes and it has added to the body of knowledge. That's good science.

 

I still consider that there is a stark contrast between good scientific method and the blurred assertions put forward to support the case for MQA.

 

It's very disappointing that Archimago's identity has been used to detract from the effort he has made. The listening test that he's pulled together deserves due credit, as does the technical analysis. It wipes the floor with the alternate empirical work put forward by the MQA proponents, which, from what I've seen, is meagre at best.

 

I look forward to a robust, evidence based refutation of Archimago's essay. If the previous comments in the negative are any guide, I'll be waiting a good while before I see much of value in that regard.

 

We'll see.

[adamdea]

2 hours ago, Fokus said:

 

That is generally correct, although many years ago I found a very old Philips document in which their 14 bit 4 x approach was named 'upsampling'.

 

To all interested: in the context of consumer audio, oversampling and upsampling always mean (except perhaps in a few freak products) increasing the sample rate of a given digital signal, accompanied with digital-domain anti-imaging/reconstruction/interpolation filtering.

 

The in/out sample rate ratio may be integer or not, in the grand scheme of things that is but a detail.

 

All other 'definitions' such as 'increasing the clock rate' or 'inserting zeroes' are obfuscating and thus useless.

Hi that's what I thought.

I suspect that most attempts distinguish over and upsampling are based on some view that this or that process (or product) is better. The especially seems to crop up with arguments that integer multiples are more natural becasue the preserve the original data or something. (a bit mystical IMHO, but possible I suppose that they are easier to do accurately)

I think the water may have been muddied further by ASRC.

[adamdea]

9 hours ago, Doug Schneider said:

 

I actually found two bits of information about this from Thierry Heeb (he was with Anagram): "Oversampling is an upsampling process where the ratio between output sampling frequency and input sampling frequency is an integer larger than 1. Upsampling is any kind of transformation providing an output sampling frequency that is higher than the input sampling frequency and not necessarily a ratio."

 

He also wrote an AES paper in 2006 on the subject, but I can't find it at the moment.

 

I also found this Spanish-language article that references Heeb and attempts to distinguish the two: http://www.revistacec.com/didactica/3014-upsampling-y-oversampling-3014.html

Doug
SoundStage!

 

Thanks Doug

It seems that different people use the terms in different ways. I suspect that at the root of this is the argument that it's somehow ok as long as you are just adding extra values, but suspect if all your samples are different. I'm not sure this is a sound engineering distinction. The distinction seems to reflect points people would like to make rather than reflecting actual usage. 

[adamdea]

45 minutes ago, astromo said:

Thanks for shedding light on the back story.

 

I raised the Moran and Meyer work a few posts back to contrast with the lack of transparency and empirical validation being offered by the proponents of MQA. It was pointed out that the M&M work was flawed.

 

I don't see that as a problem.

 

For me, that's still a positive. They published their method, theories and results and in time it was critiqued. To their credit, M&M appear to have acknowledged their shortcomings. Other researchers have gained benefit from those mistakes and it has added to the body of knowledge. That's good science.

 

I still consider that there is a stark contrast between good scientific method and the blurred assertions put forward to support the case for MQA.

 

It's very disappointing that Archimago's identity has been used to detract from the effort he has made. The listening test that he's pulled together deserves due credit, as does the technical analysis. It wipes the floor with the alternate empirical work put forward by the MQA proponents, which, from what I've seen, is meagre at best.

 

I look forward to a robust, evidence based refutation of Archimago's essay. If the previous comments in the negative are any guide, I'll be waiting a good while before I see much of value in that regard.

 

We'll see.

The major flaw with Meyer and Moran was that it turned out some of the sacds were upsampled redbook and no one had noticed. IIRC people were just asked to bring their favourite sacds. So not really a great point for hi res proponents, if methodologically a definite flaw in the study.

[Fokus]

19 minutes ago, adamdea said:

The major flaw with Meyer and Moran

 

Another flaw, IIRC, was that they had people drive for hours (!), to an alien listening room (!), with an alien system (!), judge the sound in a fairly short time, and then drive back (possibly past the pub).

 

I know what my ears are worth after two hours in a car ...

 

[MikeJazz]

On 08/03/2018 at 3:28 PM, botrytis said:

 

Don't forget, Linn also produces music and has a very fine label for it. I think that is WHY they were adamant about it. 

 

 

Linn as been adamant to the use of open codecs since the beginning...

[Thuaveta]

On 12/03/2018 at 12:28 AM, John_Atkinson said:

 

With respect, yes, this is a core belief of mine, and has been since I worked in a research lab at the end of the 1960s. You have something relevant o say, hang it on the peg of your actual name. Just as I have done all my life - and even Doug Schneider has done! Until then, while I have read your article, and certainly have opinions on what you have written, I shall keep my thoughts to myself.

 

John Atkinson

Editor, Stereophile

 

 

With respect, I generally trust the quality of the reporting in The Economist more than I trust Stereophile, as do I have more faith in The Economists' integrity and independence. Yet...

[plissken]

4 hours ago, astromo said:

It's very disappointing that Archimago's identity has been used to detract from the effort he has made.

 

I don't get that part either. The E=MC2 could have been attributed anonymously and it's significance not diluted one single iota as it relates to the anonymous donor of the theory. 

[beetlemania]

12 hours ago, Kal Rubinson said:

The only reason I posted about this at all was that I had always thought his identity was obvious to me.

 

Is this Stereophile's new M.O.? Arrive at an opinion that the average person probably wouldn't, then double-down without evidence?

[james45974]

4 hours ago, astromo said:

It's very disappointing that Archimago's identity has been used to detract from the effort he has made.

I don't think JA has a clue about how to respond, thus the try at misdirection.

 

JA, eyes forward, FOCUS!  The subject is MQA!

 

Unfortunately for JA the patriarchal world of the print magazine is fast becoming a thing of the past.  Consumers don't have to be satisfied with the magazines product selection and editorial opinion, there are many more options.  Take a look at the number of information outlets on Daily Audiophile.  With the traditional limitations inherit in print media they are in a no win situation compared to the speed and flexibility of the internet.  We are now in a world where interested consumers and hobbyists such as Archimago and Mansr have access to platforms that rival and surpass that of traditional print media.  And I find their work good for the industry, whether they use a pseudonym or not!  Hearing from those that have been personally testing MQA, such as Miguelito, is also of good value.  Keep it up!  You are doing much more in regards to MQA than the old pint boys!  JA, please try to get with the program!

 

Thanks Chris also for providing the forum for this discussion!

[Archimago]

9 hours ago, motberg said:

I do not understand your comment above....

 http://archimago.blogspot.com/

I would think this generates revenues?

 

 

So as to make sure there are no misconceptions, I do not own Madrona Digital (mentioned earlier and edited) - that is Amir of Audio Science Review.

 

Yeah, the blog does generate a few bucks in revenue from some Amazon links basically and Google AdSense. It helps buy a few CDs and inexpensive gadgets :-). Let's just say the income is nothing compared to the day job...

 

[wdw]

Perhaps Mr. JA is so completely bound up in layers of NDA and cannot easily participate.

[beetlemania]

3 minutes ago, wdw said:

Perhaps Mr. JA is so completely bound up in layers of NDA and cannot easily participate.

Could be it! He can't disclose that he can't disclose. Double-secret NDA.

[mav52]

18 minutes ago, wdw said:

Perhaps Mr. JA is so completely bound up in layers of NDA and cannot easily participate.

Don't see how, he's not a manufacturer of a product, he's the press but like a lot of press lately, fake news.   I'm just disappointed that Stereophile has decided to attack the man, rather than attack the data in Archimago's report.  My old boss at Bell Labs always told my team, you want to prove the man wrong, prove his data is wrong.

[crenca]

24 minutes ago, wdw said:

Perhaps Mr. JA is so completely bound up in layers of NDA and cannot easily participate.

 

Yep, I suspect the use of IP/NDA in this whole MQA episode is part of how these trade publications and even manufacturers got hoodwinked.  They simply did not have the expertise and experience to know what they were wading into, and the consequences there of.

[Kal Rubinson]

2 hours ago, beetlemania said:

 

Is this Stereophile's new M.O.? Arrive at an opinion that the average person probably wouldn't, then double-down without evidence?

Since when are conversations of an individual's personal recollections subject to scientific scrutiny?  If you want to believe my memory is faulty, so be it. 

[crenca]

9 minutes ago, Kal Rubinson said:

Since when are conversations of an individual's personal recollections subject to scientific scrutiny?  If you want to believe my memory is faulty, so be it. 

 

Well, since Aristotle at least.  Besides, your changing the subject.  MQA is just software, math.  What evidence do you have that it's "deblurring" marketing claims have any truth to them?  If you don't have any, that is fine - or if you do but can't discuss them (as a signator to MQA's NDA), that is fine but perhaps you can tell us about your NDA commitment.  If you do have evidence, then show us the math.

[beetlemania]

8 minutes ago, Kal Rubinson said:

Since when are conversations of an individual's personal recollections subject to scientific scrutiny?  If you want to believe my memory is faulty, so be it. 

 

You wrote that you knew Tellig was a pseudonym before you were associated with Stereophile but provided no source that informed that knowledge - falsely, as far as I can tell, claiming "it was always in the magazine" . I've been reading Stereophile for ~30 years and have known about Tellig/Gillett for several years but I think I read about it on AudioAsylum, certainly not within the pages of Stereophile. I find it curious it was "obvious" to you.