Article: A Comprehensive Q&A With MQA's Bob Stuart

[fritzg]

Where is there MQA music for sale right now?

[master]

I think you are being a bit too pessimistic. The Tidal hi-fi streaming subscriptions are apparently at about 1.5 million already. That's pretty good growth for the time it's been around. How many high-end audio manufacturers have sold 1.5 million units of anything?

I assume Tidal and streaming in general will continue to grow. If Tidal succeeds, they won't remain alone as the only company streaming better than mp3 quality.

 

Don't get me wrong. I do pay for both Spotify and Tidal and want to see them survive and even thrive. But for them to do that they need to be making money, which ain't happening. Spotify with its claimed subscriber base of 25 million still does not make a profit. Of course the studios do, but we need the likes of Spotify and Tidal to make the profit.

 

From what I've seen, Tidal stands to save quite a bit of money by streaming smaller MQA files versus what they are streaming today in the hi-fi stream. I'd guess that's the real appeal for them.

 

Most of that savings or maybe all of it will go back to MQA (or maybe the studios), which again does not help Tidal.

 

Anyways I'm just asking cause I don't see the numbers. It's good to debate about SQ, and new DACs, etc. but as things stand I don't see any money for MQA or the streaming services like Tidal and Spotify. In that context everything seems like a lot of song and dance over nothing.

 

Maybe we are not seeing it... but maybe MQA does really have a workable plan to make some $$$ for all of their troubles.

[firedog]

Posted by crenca:

Correct me if I am wrong, but I thought MQA files (of any original sample rate - including 16/44) were larger than the same file in 16/44 and FLAC compressed (what Tidal is streaming today as their "Hi-Fi" stream)?? Also, remind me again but MQA does not compress in FLAC format very well. In other words, a MQA Tidal stream will be larger than their existing/current "Hi-Fi" stream...right, wrong, or is it "who knows"?

 

This is what every explanation suggests, and the 2L samples support it.

 

Not exactly. John Atkinson submitted his own hi-res recordings to MQA to be "treated" and the hi-res version folded by MQA was smaller than the Redbook version in flac.

 

Apparently the final size of the file after MQA has something to do with the nature of the original content.

 

Basing everything you are concluding about MQA on a few examples at 2L is apparently misleading. Too small a sample size there.

 

Maybe take that into account before being so sure you already know everything about it.

[Sal1950]

So much of this reminds me of the pedaling of ObamaCare. "Just sign it, you can find out what's inside later" LOL Look where that got us.

[mansr]

Not exactly. John Atkinson submitted his own hi-res recordings to MQA to be "treated" and the hi-res version folded by MQA was smaller than the Redbook version in flac.

 

Can I have a copy of those files? Didn't thinks so.

 

If the original files have little high-frequency content, I'd expect the MQA encoding to compress better. That said, you can't simply compare the sizes of FLAC files unless they were encoded with the same parameters. Even the exact same source file encoded with different parameters can end up differing in size by 20% or more.

 

Apparently the final size of the file after MQA has something to do with the nature of the original content.

 

Basing everything you are concluding about MQA on a few examples at 2L is apparently misleading. Too small a sample size there.

 

Maybe take that into account before being so sure you already know everything about it.

 

If the 2L samples are so terrible, 1) why is Morten Lindberg gushing over them, and 2) why don't MQA/Meridian release some better ones?

[firedog]

Can I have a copy of those files? Didn't thinks so.

 

If the original files have little high-frequency content, I'd expect the MQA encoding to compress better. That said, you can't simply compare the sizes of FLAC files unless they were encoded with the same parameters. Even the exact same source file encoded with different parameters can end up differing in size by 20% or more.

 

 

 

If the 2L samples are so terrible, 1) why is Morten Lindberg gushing over them, and 2) why don't MQA/Meridian release some better ones?

 

Why don't you ask John Atkinson for copies? They are his. Why are you implying there is something nefarious about the fact that for testing purposes he sent his own recordings to MQA to be treated? Sounds like a good idea to me. What better way for him to test and listen to MQA?

 

No one said they were terrible. I tried to make the point that they are a very small sample and you might want to reign in your conclusions based solely on them.

 

You seem to be very willing to make very broad conclusions based on a handful of files from 2L. Let's see what happens when MQA material begins to get released in earnest.

 

"Why don't MQA..."?... Maybe because they are launching a new product and their schedule and priorities don't match what you think they "should" be doing. They might have different ideas than you about how they should run their business.

 

Imagine that.

[Jud]

The MQA patents are exactly useless for learning anything about the technology, let alone evaluating its performance.

 

I suppose, being non-technical, I may be too optimistic about how much can be learned from the language of the patents (Patent WO2014108677A1 - Digital encapsulation of audio signals - Google Patents and Patent WO2013186561A2 - Doubly compatible lossless audio bandwidth extension - Google Patents). Still, this guy's take was pretty interesting to me: https://mrapodizer.wordpress.com/2015/01/13/mqa-what-is-meridian-hiding/

[mansr]

I suppose, being non-technical, I may be too optimistic about how much can be learned from the language of the patents (Patent WO2014108677A1 - Digital encapsulation of audio signals - Google Patents and Patent WO2013186561A2 - Doubly compatible lossless audio bandwidth extension - Google Patents). Still, this guy's take was pretty interesting to me: https://mrapodizer.wordpress.com/2015/01/13/mqa-what-is-meridian-hiding/

 

Those patents don't reveal anything beyond what has been written in Stereophile. The only difference is they are much more obfuscated, as patents tend to be.

[Jud]

From what I've seen so far, I wouldn't be surprised if the MQA filters applied a little EQ known to be perceived by many as "better" in addition to whatever actual "de-blurring" it performs. Especially the talk from some about improved bass has me suspicious. Any "temporal blur" present in a 192 kHz recording really doesn't have any impact on frequencies normally considered as bass. Either these reviewers are imagining things, or MQA is being less than honest with the evaluation samples they provide, neither of which is unlikely.

 

Perhaps the "enhanced bass" effect comes from something prosaic rather than sinister. Remember that people (with the exception of a very few with technical knowledge) don't really have much language for describing sounds other than in the very common terms of loudness and frequency response. I think this may be particularly true of digital audio. How many people know what jitter sounds like, for example? I frankly don't.

 

So it may be that the "enhanced bass" has more to do with listeners' abilities to adequately describe what they are hearing than the actual frequency response of the MQA'd files.

 

Bearing this in mind, here's an interesting five year old blog quote about Meridian's digital filtering long before the days of MQA, describing their minimum phase "apodizing" filter:

 

Every frequency component of the sound will get a different shift in time with this filter, it is like smearing the sound over the frequency range, with lows first and highs last. This might also explain why the lows are more pronounced.

[Emphasis added.]

 

It's talking about the characteristic dispersive response (group delay) of a minimum phase filter. Such filters have also been described by listeners as providing an effect of greater depth to the sound.

 

This raises a key aspect of evaluations of sound quality in general. Particularly when long term listening (at least weeks) hasn't been done, it can be virtually impossible to distinguish "different" from "better." Does *everything* have lots of bass and depth with MQA, or does it only bring these out to the extent they're actually present in recordings?

[mansr]

Perhaps the "enhanced bass" effect comes from something prosaic rather than sinister. Remember that people (with the exception of a very few with technical knowledge) don't really have much language for describing sounds other than in the very common terms of loudness and frequency response. I think this may be particularly true of digital audio. How many people know what jitter sounds like, for example? I frankly don't.

 

So it may be that the "enhanced bass" has more to do with listeners' abilities to adequately describe what they are hearing than the actual frequency response of the MQA'd files.

 

Bearing this in mind, here's an interesting five year old blog quote about Meridian's digital filtering long before the days of MQA, describing their minimum phase "apodizing" filter:

 

[Emphasis added.]

 

It's talking about the characteristic dispersive response (group delay) of a minimum phase filter. Such filters have also been described by listeners as providing an effect of greater depth to the sound.

 

This raises a key aspect of evaluations of sound quality in general. Particularly when long term listening (at least weeks) hasn't been done, it can be virtually impossible to distinguish "different" from "better." Does *everything* have lots of bass and depth with MQA, or does it only bring these out to the extent they're actually present in recordings?

If people don't know what they're hearing and can't describe it, what possible use are their opinions to others?

[Jud]

If people don't know what they're hearing and can't describe it, what possible use are their opinions to others?

 

In the very same way it helps your doctor when he asks you, a non-physician, to describe your symptoms.

 

It is not necessary for people to understand the precise audio or engineering causes for what they are hearing, or even, as in my previous comment, to be able to describe what they are hearing accurately and precisely. It would of course make things easier if that were the case, but if in fact it turns out to be true that perceived "better bass response" from MQA is simply the result of dispersive minimum-phase filtering, what does it matter that listeners do not give technically correct descriptions of dispersion/group delay? It is enough that they can adequately describe the "symptoms."

[mansr]

In the very same way it helps your doctor when he asks you, a non-physician, to describe your symptoms.

 

Yes, but before prescribing a treatment, a physician will perform tests and measurements to determine the actual ailment.

[crenca]

Perhaps the "enhanced bass" effect comes from something prosaic rather than sinister......Bearing this in mind, here's an interesting five year old blog quote about Meridian's digital filtering long before the days of MQA......Does *everything* have lots of bass and depth with MQA, or does it only bring these out to the extent they're actually present in recordings?

 

So it begins (actually, it started along time ago), endless opinion, speculation, and armchair guess work around a fundamental format change because what is ACTUALLY happening is forever unknowable behind a IP/NDA. Even IF you want to evaluate MQA on SQ alone, you can't do it. I can't wait until it becomes the standard!

[The Computer Audiophile]

Yes, but before prescribing a treatment, a physician will perform tests and measurements to determine the actual ailment.

Hi Mansr - Just like its hard to believe the press who can't find something that's not perfect with MQA, it's a bit hard to believe your positions if you can't admit there's at least one redeeming quality of this technology. You seem to be fighting hard, with an answer for everything. But, it just makes you seem like you have a major grudge.

[The Computer Audiophile]

So it begins (actually, it started along time ago), endless opinion, speculation, and armchair guess work around a fundamental format change because what is ACTUALLY happening is forever unknowable behind a IP/NDA. Even IF you want to evaluate MQA on SQ alone, you can't it. I can't wait until it becomes the standard!

Is it at all possible that your negative opinions are equivalent to the other opinions in that they are based on speculation?

[crenca]

Posted by crenca:

 

Not exactly. John Atkinson submitted his own hi-res recordings to MQA to be "treated" and the hi-res version folded by MQA was smaller than the Redbook version in flac.

 

Apparently the final size of the file after MQA has something to do with the nature of the original content.

 

Basing everything you are concluding about MQA on a few examples at 2L is apparently misleading. Too small a sample size there.

 

Maybe take that into account before being so sure you already know everything about it.

 

Darko says:

 

"“Babylon Sisters” 24bit/96kHz file size is halved once encoded to MQA: 131.6Mb to 73.6Mb"

 

Anyone actually have this hi-res track (wav and flac) and the 16/44 track (wav and flac)? In particular, what is the 16/44 size in flac (what Tidal currently streams) compared to the MQA 73.6?

 

This way, we can answer the question as to how much more bandwidth Tidal is going to need to stream MQA...

[crenca]

Is it at all possible that your negative opinions are equivalent to the other opinions in that they are based on speculation?

 

Nope

 

My opinions are based on what we can and do know about the IP, formats and digital ecosytems, DRM, etc. I even believe that MQA is a significant SQ improvement in most cases for most recordings. WHY it is an improvement however is forever locked behind IP and NDA and this is known to all. Even if you yourself are invited to sign the NDA (or in your hush hush conversations you have had with those who do) you can not disclose to us here what you know, and we are back into the same position - the unknown.

 

edit: Also, for those who do sign the NDA and/or talk with those who do, what do you really know? Are you able to empirically test in any way what you are being told, or are these sorts of things simply a pealing back of one curtain to reveal yet another. NDA's are in most cases simply a confidence game - anyone can still snow you with mumbo jumbo and you would be none the wiser because they have not actually given you anything of significance empirically.

[The Computer Audiophile]

Nope

 

My opinions are based on what we can and do know about the IP, formats and digital ecosytems, DRM, etc. I even believe that MQA is a significant SQ improvement in most cases for most recordings. WHY it is an improvement however is forever locked behind IP and NDA and this is known to all. Even if you yourself are invited to sign the NDA (or in your hush hush conversations you have had with those who do) you can not disclose to us here what you know, and we are back into the same position - the unknown.

I believe your wrong. Patents don't last forever as your opinion clearly states :~)

[mansr]

Darko says:

 

"“Babylon Sisters” 24bit/96kHz file size is halved once encoded to MQA: 131.6Mb to 73.6Mb"

 

Anyone actually have this hi-res track (wav and flac) and the 16/44 track (wav and flac)? In particular, what is the 16/44 size in flac (what Tidal currently streams) compared to the MQA 73.6?

 

This way, we can answer the question as to how much more bandwidth Tidal is going to need to stream MQA...

 

The duration of Babylon Sisters is 5:51 according to Internet sources. In uncompressed 16/44 this would be 62 MB. A quick torrent search turns up a FLAC copy of this track at 33.5 MB, less than half of the MQA version.

[mansr]

I believe your wrong. Patents don't last forever as your opinion clearly states :~)

 

Not forever, but beyond the useful life of most modern technology. They can also be strung out for a very long time in excess of the nominal 17-year duration. Look up "extension patent" if you don't believe me.

[crenca]

I believe your wrong. Patents don't last forever as your opinion clearly states :~)

 

True, but for our context "forever" is the time it takes MQA to become the de facto standard or create 10k comment threads about WHY it sounds better....

[crenca]

The duration of Babylon Sisters is 5:51 according to Internet sources. In uncompressed 16/44 this would be 62 MB. A quick torrent search turns up a FLAC copy of this track at 33.5 MB, less than half of the MQA version.

 

So, for Tidal to stream Hi-Res of this particular track it will need about double (or a little more) using MQA stream. OR, does the 73.6Mb MQA get compressed further in FLAC, and if so by how much? Does it approach the 33.5 of the 16/44 FLAC?

 

Given these truths, how do we explain such sentences as these:

 

"...The immediate and face-slappingly-obvious benefit of MQA to Jay-Z’s company is the ability to stream hi-res content in Redbook-sized containers..."

 

That is from Darko's otherwise excellent "An inconvenient truth: MQA sounds better!". I have read similar thoughts expressed in most articles on MQA.

 

Am I being face slap by a gnat (and thus I can't feel it) or am I simply being "negative"....

[lmitche]

I'm happy to see that the Bluesound streamer will decode MQA and forward the decoded stream to a DAC. That gives me faith about availability of a software MQA decoder in the future.

[mansr]

So, for Tidal to stream Hi-Res of this particular track it will need about double (or a little more) using MQA stream. OR, does the 73.6Mb MQA get compressed further in FLAC, and if so by how much? Does it approach the 33.5 of the 16/44 FLAC?

 

The 73.5 MB has to be already compressed. 24/44 would be 93 MB. No uncompressed PCM format comes out close to 73.5 MB.

[testikoff]

Instead of claiming that MQA version 'sounds better' than the original Hi-Res recording why not first perform an honest ABX test (not looking at DAC lights or audio driver control panel) & see if the reviewer really can tell the difference between the two formats? I did with my Explorer2 DAC using 2L test bench samples (DXD vs MQA) & failed:

foo_abx 2.0.2 report
foobar2000 v1.3.10
2016-06-01 10:40:40

File A: 2L-056_03_stereo_DXD.flac
SHA1: 1dd9a4538e579fd26d1d80ee82a8b372aab93b46
File B: 2L-056_03_stereo_DXD_mqa.flac
SHA1: a8683af0e21d1c6e8776bfd29ab52ed02b638d96

Used DSPs:
Resampler (SoX) mod2

Output:
ASIO : Meridian USB2 ASIO Driver
Crossfading: NO

10:40:40 : Test started.
10:41:20 : 00/01
10:41:53 : 00/02
10:42:24 : 01/03
10:42:56 : 01/04
10:43:28 : 01/05
10:44:16 : 01/06
10:44:23 : 01/07
10:45:10 : 02/08
10:45:38 : 03/09
10:46:24 : 04/10
10:46:24 : Test finished.

----------
Total: 4/10
Probability that you were guessing: 82.8%

-- signature --
0991ab381c45475b532fe6d207fee65e442f0edb