MQA is Vaporware

[Jud]

7 minutes ago, PeterSt said:

At the time they didn't even know this was about MQA because I fooled them by telling that this was about the "Hires" - and they could choose but had to guess first which was and was not hires.

After that I said I was sorry but I had another version prepared.

All chose for this next version which was played at the (prepared) same level.

 

The former was MQA and the latter was Redbook.

The (kind of) unanimous conclusion : See we were conned.

 

What, you gave them an A/B test and they couldn't tell a difference?  Alert the media!

[Jud]

16 minutes ago, Fitzcaraldo215 said:

Did he need to?  There is a scientific concept you may have heard of called the "speed of sound", which relates the distance traveled by sound through an elastic medium, such as air, with time.    Here is the complete scientific explanation:

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

 

Does that help?

 

 

 

From MQA:

 

"The increasing evidence of acute sensitivity to time/frequency balance and practical experiments showing deterioration in sound quality from steep filters have led Stuart and Craven to conclude that the most appropriate benchmark against which to judge the blurring in a sound reproducing system is air itself. [2]

 

"Air attenuates high frequencies and disperses transients, but in a way that is completely familiar. Can we therefore mimic this behavior to give a 'more natural' system response whose only effect would be to effectively move the listener a short but familiar distance from the source?"

 

They're saying air as a medium produces "temporal blurring," and that it's most appropriate to limit digital filters to this same amount of blurring.

 

I'm not familiar with scientific or engineering proofs of either proposition.  I'm not even sure either proposition is well formulated, i.e., sensible.  It's some scientific or engineering evidence of either that's being asked for.

[Jud]

33 minutes ago, Fitzcaraldo215 said:

Makes intuitive sense to me.  Digression:  there is a parallel concept in vision called "atmospheric perspective", where more distant objects are increasingly blurred and robbed of color.  This has been known by artists since the Renaissance, who combined geometric and atmospheric perspective in their 2D paintings to generate a perception of depth that was often uncalnny. 

 

But, what does the reference [2] you quoted say? Shouldn't that explain and provide some basis for your concern?  

 

Why are you and mansr asking us here, when deep scientific questions like this are absolutely not going to get any sensible answers from the likes of us dodos?

 

Note this reference is for something Stuart and Craven concluded, so it could be a reference to their statement of the proposition rather than evidence of it.  (I haven't looked yet.)

 

I was explaining what mansr was asking about (whether any evidence of these propositions was given as part of the MQA presentation or in handouts).  Neither of us was asking the folks here, though of course answers from whatever source are welcome, I'm sure.

[Jud]

10 hours ago, PeterSt said:

This is how the Scarlet Book specification depicts DSD (SACD) to be -6dBFS opposed to its PCM counterpart. This is not always obeyed and with e.g. -3dBFS only, modulator overload may occur (may, because it depends on the music (data) itself), that resulting in really nasty distortion.

 

Though of course correct, I can say that using SoX and HQP, both of which report overloads, I have virtually never run into any with a -3dB setting.

[Jud]

7 hours ago, Hifi Bob said:

However, filter ringing ('blurring' in BS terminology) in the domain of interest—i.e. Redbook—occurs at or very close to 22kHz.  So it does not affect sound at all (ultra-sound might be affected but by definition, it's inaudible).

 

- Some folks (not just MQA) think this affects the audible range.

 

- It has been determined that the MQA decoding filters (at least those in the first Meridian MQA DAC) are not steep enough to eliminate ringing that may exist in a recording. Therefore any elimination of ringing must occur, if at all, during encoding.

[Jud]

Two comments:

 

- I think we still have a way to go in finding out what causes us to hear a reproduction as accurate versus the original.

 

- The commercialization and confidentiality involved with various efforts that we are told aim toward that goal (for example MQA, Schiit Audio's "Manhattan Project") hinder our ability to find out if there's really anything worthwhile in them.

[Jud]

5 hours ago, Jud said:

- I think we still have a way to go in finding out what causes us to hear a reproduction as accurate versus the original.

 

- The commercialization and confidentiality involved with various efforts that we are told aim toward that goal (for example MQA, Schiit Audio's "Manhattan Project") hinder our ability to find out if there's really anything worthwhile in them.

 

In response to the second point, you said:

 

3 hours ago, The Computer Audiophile said:

- I agree for the most part. However, the rubber meets the road when we listen to music. The chefs are saying "trust me" with the ingredients, but we get to taste the food. If we don't like the food we don't buy it and we tell others that we didn't like it. 

 

Then, responding to someone else:

 

3 hours ago, The Computer Audiophile said:

I would really like to correlate measurements into sonic attributes for my own knowledge / education. If something measures excellent, but I don't like the sound, I'd like to know why. I'd never be able to look at data sheets and decide if a product is for me or not. 

 

Agreed that I'd love to be able to correlate measurements with what I hear and like.  As you say, that would make it easier to tell whether a specific product is for me.

 

That's why I'd prefer not to be limited to "tasting," great though it is.  Because I don't get to learn what ingredients made something taste good to me.

 

Does MQA sound slightly less good to me than the corresponding non-MQA hi res file because of some of the characteristics mentioned by mansr, Peter, Miska, Fokus and others?  Is it because it uses minimum phase filters, and I have always preferred linear phase filters in my system?  For those who like it, is this because they are getting some artificial "kick" from intermodulation distortion (as with some NOS DACs), or because MQA is doing something at the encoding end that allows it to handle the time domain better in their systems?  Impossible to tell for certain at this point, which is a little frustrating.

[Jud]

3 minutes ago, crenca said:

SQ assessment is simply not this radical and varied.

 

http://deutsch.ucsd.edu/psychology/pages.php?i=201

 

Quote

 

The following entries describe and illustrate some of Deutsch’s illusions of music and speech. Many of them show that people can differ strikingly in the way they hear very simple musical patterns. These disagreements do not reflect variations in musical ability or training. Even the finest musicians, on listening to the stereo illusions described here, may disagree completely as to whether a high tone is being played to their right ear or to their left. And the most expert musicians, on listening to the tritone paradox, can engage in long arguments as to whether a pattern of two tones is moving up or down in pitch.

 

 How do we explain these striking perceptual discrepancies? In the case of stereo illusions such as the Octave Illusion, the Scale Illusion, the Chromatic Illusion, the Cambiata Illusion, and the Glissando Illusion, disagreements tend to arise between righthanders and lefthanders, indicating that they reflect variations in brain organization. In contrast, the way the Tritone Paradox is perceived varies with the geographical region in which listener grew up, so differences here are related to the languages or dialects to which people are exposed.

 

 

[Jud]

15 minutes ago, The Computer Audiophile said:

If we are going back to the original performance a the gold standard, does anyone know how we can tell which version is more accurate?

 

We can't.  The only thing I've ever found to work is that you have to listen to a variety of music and see whether you're getting a wide variety of sounds (often on different tracks in the same album, or even within one track - "Good Vibrations" was recorded in four different studios to give different sounds to various parts of the song).  To the extent you are, then your system is being accurate.  To the extent there may be an overall sameness to the sound, that's coming from a sonic signature your system is imposing on the music, and is thus inaccurate.  (Barry Diament and @PeterSt have said much the same thing from time to time.)

[Jud]

3 minutes ago, crenca said:

 

 

Interesting Jud, but I find it likely that these researches overplay their hand.  It is a common problem in academia as one has to emphasize in the competition for grant money.  In my experience, the differences in SQ assessment (and they are real) are far outweighed by the similarities.

 

All this was more widely understood when people in our Western Civ had more direct experience with traditional instruments and their sound.  Now that people are trying to get an handle on studio recordings and their strong "artificial" elements, a radical subjectivity makes more sense but for reasons I already stated I don't really buy it...

 

Our brains are terrific pattern matching engines.  Think of optical illusions you can't "unsee."  These result from some combination of inborn and trained pattern recognition that overrides the actual light waves hitting the retina.  Same with audio illusions.  These are just illustrations of a larger phenomenon, which is that our genes and our histories lay down patterns we recognize, including patterns for what sounds "right." 

 

I've had Vandersteen speakers for over 25 years.  They have certain imaging characteristics that have become key to my hearing a reproduction as "right."  On the other hand, the same crossover design that allows these imaging characteristics also creates a frequency response hump.  Some people find this intolerable; I completely listen past it.

 

Whether you realize it or not, what you think sounds accurate has to do with who your parents were, the language you spoke growing up, pieces of equipment you've had in your system for a long time, and many other factors unique to you.  Or do you really think people who like equipment you don't just can't hear well? 

[Jud]

52 minutes ago, crenca said:

 

 

Nope, it does much more than that.  The experience is not nearly so relative/subjective as you making it out to be.  Besides, experience of music is never (or should not be) this monotonous.  I have experience of acustic intstruments in many different settings, room conditions, etc.  From this, I am able to identify Fidelity to a reasonable degree - I can hear through any given playback chain an level of "accuracy" of a trumpet or violin and this accuracy is transferable - others can confirm it.  Your simply denying it with a radical subjectivist take on fidelity.  

 

This shows something quite different than you think it does.

 

First tell me: Do you recognize these as the same song?  Then we'll go from there.

 

 

 

[Jud]

2 hours ago, Ralf11 said:

It is by no means clear as to whether or not someone could train themselves to accurately recall a particular violin sound over several months.

 

There have been many tests performed in front of orchestra members and classical critics where the record of picking Strads and Guarneris against modern instruments (sometimes even including carbon fiber violins) has been pretty abysmal.  These people have more experience of the sound of these instruments than anyone on this forum ever will.

[Jud]

We need names for these - the Beastie Boys principle, the Hugging Wife principle....

 

I have an idea about the Hugging Wife principle, Peter, which you may or may not like (and which may or may not be right, of course).  Later, when I have a minute.

[Jud]

26 minutes ago, lucretius said:

 

The fact that you could identify your wife's voice implies that the 8bit/32kHz recording was accurate enough for you to do so -- accuracy in these circumstances is a matter of degree, not an absolute.  The various arguments I've seen in this thread are arguments against absolute accuracy but not arguments against accuracy as a matter of degree. Surely one goal (but not necessarily the only goal) of reproduction is to increase the degree of accuracy as much as is feasible?

 

It is not even a matter of degree.  Look at my post in this thread yesterday regarding the two versions of "My Favorite Things:" https://www.computeraudiophile.com/forums/topic/30381-mqa-is-vaporware/?page=97#comment-667604 .  Listen to a bit of each version.  Now: How do you know these are two versions of the same song?

 

[Jud]

12 hours ago, The Computer Audiophile said:

 

Absolutely not. 

 

Sure you can identify a trumpet, but that has little relation to accuracy. I can identify my wife's voice sampled at 8 bit / 32 kHz, but that don't mean the recording is accurate. 

 

Ask a violinist if violins sound different. The answer will be yes. If you don't know the exact violin in use, you have no way of judging accuracy. 

 

 

http://www.thestrad.com/blind-tested-soloists-unable-to-tell-stradivarius-violins-from-modern-instruments/

 

http://www.thestrad.com/players-favour-new-violins-over-old-in-largest-ever-blind-testing-experiment/

[Jud]

1 minute ago, The Computer Audiophile said:

 

The second link bolsters my point that violins sound different and violinists identify these differences. 

 

"The final results showed that one modern instrument garnered a total of 26 points, being the top choice for four players, second choice for another four, and rejected by two. Conversely, a Stradivari ended up with a score of -9. Its closest rival was a modern instrument, which had a score of -7."

 

Yes, but why can't professional players, even when playing the instruments themselves (first link), match the pattern of a Strad?  They've surely heard the sound over and over again for years.

 

(There is an answer to this.)

[Jud]

26 minutes ago, The Computer Audiophile said:

 

Good question. Perhaps, if the Strad is the gold standard and people build to match the gold standard, they've done a great job of matching that gold standard. 

 

But if they matched the standard with the modern instruments, you wouldn't get the strong preferences indicated in the results from the second link.

[Jud]

49 minutes ago, Jud said:

 

Yes, but why can't professional players, even when playing the instruments themselves (first link), match the pattern of a Strad?  They've surely heard the sound over and over again for years.

 

(There is an answer to this.)

 

If you look at the circumstances of the test, these players were not being asked to match a pattern, but for preference:

 

Quote

The players were told to judge each violin as if they were looking for an instrument that could best replace their own for an upcoming concert tour.

 

Note that regarding *preference* there were definite answers, as stated by one of the people involved in setting up the test:

 

Quote

For me, the really important finding is that whereas soloists readily separate instruments they like from those they don’t, they seem unable to tell old Italian instruments from new ones....

 

[Jud]

44 minutes ago, Kal Rubinson said:

They share an identifiable melody and a title based on that melody but I do not regard them as the same song.  

 

How did you identify the melody as being shared?  One is sung, the other played on an instrument.  They have different rhythms.  Are they even the same notes (i.e., same key/pitch)?

[Jud]

16 minutes ago, lucretius said:

 

One has no words so it's not a song.  ????  Do you mean to ask if I consider them both the same melody?

 

You yourself have said that they are of different rhythm and possibly different pitch -- that would make them different melodies.

 

 

 

So every time someone sings or plays a song in a new key you don't recognize it at all?

[Jud]

5 minutes ago, lucretius said:

 

If the words are substantially the same, it's the same "song".  Yes, I can recognize (sometimes) when a melody is a derivative (another interpretation of a composition).

 

How do you do that?"

[Jud]

35 minutes ago, lucretius said:

 

That's a good question for which I haven't given much thought.  I assume it's pattern recognition.

 

Yup.  Now think about this: You can easily have a melody that shares many more of the same notes than do melody "A" and melody "A" in a different key, but your impression is that it's a *different* melody.  Agreed?

[Jud]

19 minutes ago, Kal Rubinson said:

Common harmonic progressions although time and pitch are shifted and variable.  Its like looking into the mirror and "seeing" my father although we differ in so many ways and are not the same people.

 

Yep.  And even if a player bends some of those notes and plays a little with those progressions, you might recognize it anyway.

 

In an analogous way, as I mentioned before, my speakers give me an impression of realism in spite of the fact that there's a "hump" in their response.  (Same with a lot of people, as the Vandersteen 2 series are the best selling high end speakers ever.)

 

You know those speaker response curves and waterfall plots that are in virtually every speaker review?  Most people are quite insensitive to that sort of overall response spectrum.

 

So getting all the way back to what Chris was talking about, one of the most common types of measurements you'll see has relatively little to do with our impression of what's real, just as having the most notes in common has little to do with our recognition of the pattern that says "Ah, it's 'My Favorite Things.'"

[Jud]

23 minutes ago, jabbr said:

If I want real accuracy I can either pay $$ to listen to a real live concert, or bring good wine over to my friends' places and listen in person ... perhaps to a Strad if I bring really good or really interesting wine

 

I want to know the wine that can get you your own chamber concert with a Strad.

[Jud]

7 hours ago, fung0 said:

it's hard to be confident that lack of demand would kill it.

 

I don't think the music industry works on the Sears model.

 

What I'd be more concerned about is whether the industry thinks there's any market for hi res streaming (or even continued download availability) outside of MQA.