Jump to content

MQA Users & Discussion Thread


Guest AndrewC

Recommended Posts

6 hours ago, legend said:

 

It is true that in the 3 cases (from 2 different recording companies) where I have been able to compare the same recording in different formats I have so far preferred MQA - but am still looking for other cases.

 

But although I often do, I don't always enjoy recordings in MQA on Tidal.  A few days while searching on Tidal for some classical piano music (always a good test for fidelity) I came across the some Chopin Etudes played by Maurizio Pollini from the 1970s - and they sounded so awful (hard, bright, jangly etc) that I quickly stopped playing.  I have an LP of the Chopin Polonaisen & Nocturns played by the same artist that won the "Grande Prix du Disque" in the 1970s and it sounds fine!

Sorry if this is off topic due to not mentioning MQA, but Rod, I just confirmed today that Pollini's 'Etudes' CD is also of hard, compressed, headache-inducing quality. It was released in 1972; all the other CDs in his Chopin set were recorded from 1975 to 2008, and sound progressively better. However, they all seem to have an uncomfortable DG bloom, suggestive of reverb, and a muddiness from about middle C downward.

 

The best quality piano playing (in musicianship, disappearance of 'technique', superb voicing, etc) and piano recording quality I've come across, is the 24/96 FLAC version of this from 2017:

https://www.prestoclassical.co.uk/classical/products/8303882--volodos-plays-brahms

 

Might be helpful for somebody? You can get a gist from even the samples.

  • Like 1
Link to comment
Share on other sites



19 minutes ago, eltech said:

Impartiality is something that can be learnt

For example, I have a feeling a relationship councilor or mediator would be able to sift through emotive language and remain impartial to assist their clients. To suggest it can't be done in the field of audio evaluation seems far fetched when many professionals in various fields do it daily.

Link to comment
Share on other sites

Guest Eggcup The Daft
2 hours ago, eltech said:

Different typologies, and monotonicity.

I sort of expected a longer answer here. I thought that delta sigma DACs were inherently monotonic, and so are all CMOS chips due to the nature of the design used. So why is it relevant today?

I presume I'm missing something obvious here.

4 hours ago, legend said:

 

I guess because any MQA distortion is still well below speaker distortion (even at 0.1%) then it is not apparent (to me at least) - but it will inspire me to try to get the speaker distortion even lower!

And now i'm really going to show my ignorance. I picked up the idea somewhere along the way that you can hear smaller differences in distortion of source components through loudspeakers with relatively high distortion. Is that not so?

Link to comment
Share on other sites

33 minutes ago, Eggcup The Daft said:

sort of expected a longer answer here

You were asking about multibit DACs. What did you want me to say?

33 minutes ago, Eggcup The Daft said:

So why is it relevant today?

I presume I'm missing something obvious here

Edited by eltech
Link to comment
Share on other sites



  • Volunteer
13 hours ago, legend said:

 

I guess because any MQA distortion is still well below speaker distortion (even at 0.1%) then it is not apparent (to me at least) - but it will inspire me to try to get the speaker distortion even lower!

The logical consequence of your comment is that pretty much nothing matters apart from the speakers. Just about any well designed component will have lower distortion than the speakers and therefore by your argument differences between components will be inaudible.

Is that really what you mean?

  

Link to comment
Share on other sites

  • Volunteer
10 hours ago, eltech said:

I don't know how to answer this. I agree that we all have biases and a blind test would be the most ideal. I was really just saying I'm not the average forum punter with a home stereo. Which is not to say that the average punter's observations aren't valid, just that when you have to listen professionally it's a different game. 

 

Impartiality is something that can be learnt.

This is a very interesting article on bias

https://www.newyorker.com/tech/frontal-cortex/why-smart-people-are-stupid

 

Perhaps our most dangerous bias is that we naturally assume that everyone else is more susceptible to thinking errors, a tendency known as the “bias blind spot.” This “meta-bias” is rooted in our ability to spot systematic mistakes in the decisions of others—we excel at noticing the flaws of friends—and inability to spot those same mistakes in ourselves. Although the bias blind spot itself isn’t a new concept, West’s latest paper demonstrates that it applies to every single bias under consideration, from anchoring to so-called “framing effects.” In each instance, we readily forgive our own minds but look harshly upon the minds of other people.

 

Link to comment
Share on other sites

11 hours ago, eltech said:

I don't think sight should invalidate honest descriptions of listening tests.

Yet, specialists in that field do.  Knowledge about the test can easily skew the results.   In fact, far from being a possibility, knowledge about the test almost certainly skews the results.

 

11 hours ago, eltech said:

do actually have the ability to differentiate small differences, and remain impartial and unbiased.

Just because the former is true ... doesn't mean that the later is not a real problem.

 

11 hours ago, eltech said:

not interested in brands or labels on a file. I make my judgements

 

The problem is not usually as obvious as 'wanting one to win', or being biased to a brand, or technology...... and can be a simple as if you are expecting to find a difference between A and B, then you will.

 

Testing which introduces basic controls  (eg.  A and B will be occasionally the same) can really show this up.

 

 

 

  • Like 1
Link to comment
Share on other sites

12 hours ago, BioBrian said:

Sorry if this is off topic due to not mentioning MQA, but Rod, I just confirmed today that Pollini's 'Etudes' CD is also of hard, compressed, headache-inducing quality. It was released in 1972; all the other CDs in his Chopin set were recorded from 1975 to 2008, and sound progressively better. However, they all seem to have an uncomfortable DG bloom, suggestive of reverb, and a muddiness from about middle C downward.

 

The best quality piano playing (in musicianship, disappearance of 'technique', superb voicing, etc) and piano recording quality I've come across, is the 24/96 FLAC version of this from 2017:

https://www.prestoclassical.co.uk/classical/products/8303882--volodos-plays-brahms

 

Might be helpful for somebody? You can get a gist from even the samples.

 

Many thanks for the info and link - will download the latter as I am always on the lookout for good piano recordings as tests for my speakers.

 

It sounds as if the head-ache inducing quality I heard with the Etudes on Tidal Masters is therefore not the 'fault' of MQA but also MQA has been unable to correct it - can't make a silk purse out of a sow's ear!

 

BTW the Pollini Polonaises etc that I have on vinyl was recorded even earlier in 1968 but by EMI and not DG whose classical recordings I often find 'harsh' as I also often find 2L and Linn - must be the northern climes and those harsh winters!

https://www.discogs.com/Maurizio-Pollini-Chopin-Polonaisen-Nocturnes-Ballade/release/5619923

  • Like 1
Link to comment
Share on other sites



1 hour ago, davewantsmoore said:

The problem is not usually as obvious as 'wanting one to win', or being biased to a brand, or technology...... and can be a simple as if you are expecting to find a difference between A and B, then you will.

But what if a person is expecting nothing. They have no expectations. Then what?

 

What if there is no prior suggestion that something should sound different?

 

Perhaps there is no question asked at all, rather, simply a perfunctory act of listening.

 

Are you suggesting people can't perform perfunctory acts? 

 

Why does the word exist if not to describe a human trait?

 

 

 

 

 

Link to comment
Share on other sites

2 hours ago, davewantsmoore said:

if you are expecting to find a difference between A and B, then you will.

Sure. So to be unbiased, just listen without the expectation

Then you remove the "expectation bias"

 

It is possible to not expect things. 

  • Like 1
Link to comment
Share on other sites

10 hours ago, davewantsmoore said:

Knowledge about the test can easily skew the results.   In fact, far from being a possibility, knowledge about the test almost certainly skews the results.

Well I presume you're talking about my observations regarding the upsampled files.

I approached them with curiousity, but without expectation.

It was the first time I'd knowingly heard a min-phase filter. (the file was named min-phase FYI)

I didnt care one way or the other which sounded best it was just a bit of fun.

Furthermore, If those upsampled files sounded better to me, I'd be asking the person how to do it, and I'd convert all my files.

I would like just the same as everyone else to get better sound. Who wouldnt want that? especially since I guess its free to upsample with some software.

 

 

 

Edited by eltech
Link to comment
Share on other sites

3 hours ago, Sir Sanders Zingmore said:

The logical consequence of your comment is that pretty much nothing matters apart from the speakers. Just about any well designed component will have lower distortion than the speakers and therefore by your argument differences between components will be inaudible.

Is that really what you mean?

  

 

I was being facetious - partly because I had run out of time last night and partly because I find trying to always ‘win’ arguments for their own sake a bit tedious.

 

The real reason is much the same as Eltech’s – because I spend much of my time while designing speakers listening to changes in components etc but always comparing the sound to live acoustic concerts that I attend as frequently as possible.  

 

People will probably find this ‘elitist’ – but is not really much different to wine, art etc judges where experts in their fields do have some advantages in making their judgements.  Or even doctors or dentists etc whose judgements I have to trust.

 

I don’t hear any obvious differences (which I would regard as subjective distortion) caused by MQA compared to a live event, however as a scientist I am also acutely aware that one can be wrong and am constantly rechecking my judgements – and that others may differ!

  • Like 1
Link to comment
Share on other sites



3 hours ago, Sir Sanders Zingmore said:

This is a very interesting article on bias

https://www.newyorker.com/tech/frontal-cortex/why-smart-people-are-stupid

 

Perhaps our most dangerous bias is that we naturally assume that everyone else is more susceptible to thinking errors, a tendency known as the “bias blind spot.” This “meta-bias” is rooted in our ability to spot systematic mistakes in the decisions of others—we excel at noticing the flaws of friends—and inability to spot those same mistakes in ourselves. Although the bias blind spot itself isn’t a new concept, West’s latest paper demonstrates that it applies to every single bias under consideration, from anchoring to so-called “framing effects.” In each instance, we readily forgive our own minds but look harshly upon the minds of other people.

 

That was an interesting article. I've read a few similar to that.

 

Here are a couple of articles I like, because they ask the reader to consider things from a different point of view. Perhaps after reading them you might hopefully understand what I was talking about when I say it is possible to learn to be unbiased and without expectation.

 

https://personaltao.com/taoism/truth/

 

http://www.zenthinking.net/blog/abandon-the-wanting-mind

 

 

  • Like 1
Link to comment
Share on other sites

59 minutes ago, eltech said:

Well I presume you're talking about my observations regarding the upsampled files.

More speaking in general....  it's advice I would give anybody.

 

1 hour ago, eltech said:

But what if a person is expecting nothing. They have no expectations. Then what?

Simply knowing that A and B are different things, leads to some sort of 'expectation'.

 

1 hour ago, eltech said:

Sure. So to be unbiased, just listen without the expectation

Then you remove the "expectation bias"

 

It is possible to not expect things. 

More difficult than it sounds (understatement) to simply "turn off" such a thing.    Good testing design will ensure such a thing is removed from the equation.

 

Results changing when controls are introduced strongly suggests things about uncontrolled testing.    I spent a long time trying to use my 'experienced' musician, speaker designer, music lover, etc. ear ...... and as I started to control the results more, I started to encounter all the things said about subjective testing / control.    Alas it is all quite true.

 

 

I'm not posting this to make the point "your results are invalid" ... as much as I am trying to say, if you are interested enough in this (to actually do any testing - as few people are), then you might be interested to introduce controls, and see how you go.

 

Link to comment
Share on other sites

33 minutes ago, davewantsmoore said:

if you are interested enough in this (to actually do any testing - as few people are), then you might be interested to introduce controls, and see how you go.

Sure. Thanks for your advice.

Link to comment
Share on other sites

  • Volunteer
1 hour ago, eltech said:

That was an interesting article. I've read a few similar to that.

 

Here are a couple of articles I like, because they ask the reader to consider things from a different point of view. Perhaps after reading them you might hopefully understand what I was talking about when I say it is possible to learn to be unbiased and without expectation.

 

https://personaltao.com/taoism/truth/

 

http://www.zenthinking.net/blog/abandon-the-wanting-mind

 

 

Thanks for the links.

For some reason I can't open them at work but I'll have a look when I get home

  • Like 1
Link to comment
Share on other sites

Guest rmpfyf
17 hours ago, LHC said:

However a lot of discussion in these forums suggest that it is a good idea to do some blind tests every now and then just to verify that bias has not take hold somehow. 

 

Archimago tried just this much a while ago:

http://archimago.blogspot.com.au/2015/04/internet-blind-test-linear-vs-minimum.html

 

Results were interesting:

http://archimago.blogspot.com.au/2015/07/the-linear-vs-minimum-phase-upsampling.html

 

Interestingly these were the same filters used in the files @eltech was listening to. The differences in those filters are large and IMHO audible; the redistribution of spectral energy in time is extreme. It's of course possible to tune the filter characteristics to somewhere in-between or a different altogether. 

Link to comment
Share on other sites



16 hours ago, rmpfyf said:

 

Archimago tried just this much a while ago:

http://archimago.blogspot.com.au/2015/04/internet-blind-test-linear-vs-minimum.html

 

Results were interesting:

http://archimago.blogspot.com.au/2015/07/the-linear-vs-minimum-phase-upsampling.html

 

Interestingly these were the same filters used in the files @eltech was listening to. The differences in those filters are large and IMHO audible; the redistribution of spectral energy in time is extreme. It's of course possible to tune the filter characteristics to somewhere in-between or a different altogether. 

Another interesting archimago article that one. However none of the results showed any statistical significance so either the power of the study was inadequate with too few respondents, or there was no actual statistical difference. So no conclusion can be drawn from those internet results. Oh well.

Link to comment
Share on other sites

Guest rmpfyf
2 hours ago, Ittaku said:

Another interesting archimago article that one. However none of the results showed any statistical significance so either the power of the study was inadequate with too few respondents, or there was no actual statistical difference. So no conclusion can be drawn from those internet results. Oh well.

 

That's not quite true. It's statistically significant that most listeners found the upsampled files different, which is (a) pretty good for a completely voluntary study someone's offering off their free time and (b) consistent with what's been observed for distribution formats built around minimum phase filtering (e.g. MQA).

 

Now... whether or not you like that difference or not is a different matter, and the current data is inconclusive. I would suspect there's so many biases in the 'liking' a bit of pre-ring or lesser immediate energy and a long post-ring that it'd be a very long study at least, and a rig allowing selective filtering is probably the only thing that's going to shut everyone up most of the time. 

 

It's fair to say that we've grown up on a diet of symmetric, linear filters whether digital (Redbook etc) or analogue (the effective mechanical response of a better vinyl rig tends towards linear filter response, with shorter responses in the time domain - much like higher-frequency material). Minimum phase is different. There's an interesting psychoacoustic argument as to whether that's better, more or less pleasing or whatever.

 

But unless there's a cradle-to-grave mastering process it's really just distortion. You can't 'MQA' a track to make it 'better', convolving of that nature just redistributes spectral energy to give a time-domain response that's not what the production engineer signed off on as a finished, complete work of art. Same goes with upsampling - it might certainly, subjectively be pleasant under some circumstances, it's only definitively different.

 

Whether MQA has a licensing model that allows a cradle-to-grave min-phase or not-your-average-linear-phase filter process to penetrate the market in any meaningful way or not is TBD. I'd suggest it unlikely. Their primary business is not with content creators, not least as this much is something any other PCM format can handle with the same or greater fidelity.

Link to comment
Share on other sites

I have just come across this assessment of MQA by Steven Stone who is a recording engineer by profession but occasionally writes for the Absolute Sound Magazine (so some will not trust but I have found over the years a non BS artist) as their digital expert:

 

“Since MQA was announced I’ve attended more than a half-dozen public demos in addition to the many hours I’ve spent listening at home on my own systems. At AXPONA 2018 I had a chance to partake of another demo, courtesy of Peter McGrath in the Paragon Audio room, with the Wilson Audio Alexandria XLF loudspeakers. Actually, I attended the demo twice. The first time I sat in the prime listening seat while during the second session I sat in the seat directly next to the prime listening seat. McGrath’s demo, using his own stunningly good recordings, codified for me two vital details about MQA demos and MQA listening in general: My first truth, which I have found universal, is that if you really want to hear what MQA can do, you must begin with a well-recorded phase-correct recording of instruments in a real acoustic space. MQA can’t make a badly mixed studio recording sound good. My own live concert recordings with the purist sonic characteristics have sounded better when encoded into MQA. Peter McGrath’s magnificent recordings were also improved by allowing an easier listen into the mix and more readily identifiable soundstaging. Peter called it “more human.” To hear MQA’s subtle but powerful differences you must listen to a good recording of Mahler or Mozart, not Metallica.

The second and equally important detail about MQA A/B listening comparison tests is if you are not in the prime listening position, which is the one that triangulates most precisely with the transducers used during the listening session, you are not in a position to judge MQA sound quality properly. When I was even one seat to the side of the central listening position the effects of MQA were vastly reduced to the point where they were almost nonexistent. I have been in sessions where even professional recording engineers (and audio reviewers) have made dubious (and IMHO downright wrong) assessments of MQA’s sonic effects based on a group session where almost everyone in the room is in a bad seat (which is any seat that is not properly triangulated with the transducers). In all the other seats you might as well be listening through a Dixie cup and a string—really.

So, if you have an opportunity to participate in a group listening session evaluating MQA, unless you have THE SEAT, i.e. the sweet spot, your opinions are not going to be correct and you won’t be hearing any of what I have come to believe are fundamentally profound (but subtle) sonic improvements that MQA-encoded files can have over the original WAV (or even DSD) masters. Let the flames begin…”

 

It suggests to me that some of the perceived (by me and some others!) benefits of MQA are due to its timing effects (creating a more realistic acoustic space etc) - and that

1.  DBT of MQA is fraught to do en masse  - choice of music and location of listeners etc.

2.  headphones may avoid the location problem - one's brain is always then in the central listening position.

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

  • Volunteer
29 minutes ago, legend said:

occasionally writes for the Absolute Sound Magazine (so some will not trust but I have found over the years a non BS artist) as their digital expert:

TAS has probably been worse than Stereophile in terms of destroying the credibility of audio reviewers. 

Have a look at what the fawning sycophant Robert Harley writes:

https://mqa-production.s3.amazonaws.com/default/0001/01/6bbf4eddcd5866f75cc5219898f0c7944817edb3.pdf

 

so yes, I’m afraid that anyone writing about MQA in that magazine must have their credibility called into doubt. 

Link to comment
Share on other sites



  • Recently Browsing   0 members

    • No registered users viewing this page.




×
×
  • Create New...
To Top