MQA Deep Dive - I published music on tidal to test MQA

[JSmith]

Amir from Audio Pseudoscience Review strikes again..

JSmith

 

[Lunafag]

if it sounds good to you than it's good for you.

There are codecs that achieve this for a fraction of the filesize and aren't trying to squeeze money out of every step of audio production and consumption chain while using underhanded tactics to do so.

 

[pjug]

You can make that determination on your own,,me personally I don't need Amir or John Yang to make that for me, people get caught up in the codecs trying to over analyse without listening to the music,if it sounds good to you than it's good for you.

Reminds me of the Bobby Knight "relax and enjoy it".

 

[Jimbob54]

Here is a thought exercise. Let's just mentally park the folding/ filter side of MQA (the supposedly clever bit) and imagine instead the following (imaginary) scenario:

Tidal was the same as Qobuz- both offering up to "hi-res" PCM that could be played on all DACs that can go up to 192.

Then one day, Tidal re-jigged their tiers- lossy (as it is now), redbook (so proper "lossless") and Hi-res (everything above 16/44 or 16/48) BUT- to access Hi Res you had to buy a "Tidal Hi res" certified DAC.

I know I have my own ideas on how successful that move would be.

The only reason I can see that MQA still exists at all is that Tidal allow the core decode in their player apps. Their new tier (which I assume is being trialled on Australia) might actually be the death of MQA- because if people dont pony up for the new "master" tier , rather than it being bundled with the "hifi" tier as was, there will be even fewer people listening to MQA and even less appetite from the labels to support it.

 

[muslhead]

Nonsense. @amirm is no leader (well, possibly to the mods, but not otherwise). He's just the gracious host of this forum. Why would you expect him to take a position because of that? What would he taking a position bring to the table that other users can't? Amir can have any position he likes, even an ambivalent one, just as anyone else.

And because we have no leader we have threads like this.

 

[Raindog123]

MQA encoding purposefully reduces the fidelity of the digital music data you put into it, and it does so for two reasons. One of those reasons - to preserve ultrasonics - you know is pointless since we don't hear those frequencies. The other reason - to "deblur" the music by applying slow, apodizing filters - you know is pointless at best and damaging at worst, since you have noted in countless reviews the problems with slow/leaky filters.

So the best-case scenario is that MQA files sound the same as PCM files. But even in that best-case scenario, MQA files use lossy encoding in order to achieve signal-processing goals that you yourself have repeatedly said are pointless and/or undesirable

And throw in that (a) there is no proof that even a triangle-shaped source material is lossless. It would be very easy for MQA to run such test case and publish the results. And yet we don't have it. And (b) the “deblur” or any superiority (as in ‘better than lossless’) is very easy to validate through controlled listening tests. With again, no such proof existing/offered.

Which again at this late point in the game is a strong indication of them have tried and failed.

 

[ebslo]

Not at all. There is for example "perceptually lossless." This is when you tell a lossy encoder to produce the highest quality it knows, and have the bit rate vary. This is as special VBR mode that is available for some lossy codecs. We had this at Microsoft and in vast amount of content it was transparent and about half the size of mathematical lossless.

MQA proposes another form of lossless that says it music follows statistical distribution, they are able to fully preserve its actual informational content including what is in ultrasonic. If they are able to achieve this fully, what is your beef with it? That it doesn't spit out a bunch of bits for a rectangular channel is not used in music?

When we want to say the same bits come out that go in, we clarify with the term "mathematically lossless."

Remember, there is no lossless codec that works for all content. Here is the wiki on that: https://en.wikipedia.org/wiki/Lossless_compression

"By operation of the pigeonhole principle, no lossless compression algorithm can efficiently compress all possible data. For this reason, many different algorithms exist that are designed either with a specific type of input data in mind or with specific assumptions about what kinds of redundancy the uncompressed data are likely to contain."

The code books in lossless codecs for example are trained on a specific dataset which in this case would be music, not random test signals. Lossless codecs can actually make the output larger than input! When this happens, they cheat and just pass the input to output. MQA can't do that because it explodes the output so it complains with error messages.

Could you please clarify the apparent disparity between the above and your prior post on lossy vs. lossless?

Of course. The definition of a lossless codec is that it can mathematically be converted back to the source with not a single bit out of place. MQA does not at all qualify for this. I remember going to first audio show where the MQA team had just rolled out the format and them asking me, "do you know what MQA is?" I said, "yes, it is a lossy format to encode high-res music." They were horrified at me saying this but this is what it is. With flac encoding, the baseband is already very near where information theory predicts. You can't pile on the full ultrasonic spectrum on top of that.

 

[JSmith]

Could you please clarify the apparent disparity between the above and your prior post on lossy vs. lossless?

There is no disparity... it's just you may not understand the content and concept of the replies by Amir. I can't see how it could be explained clearer.



JSmith

 

[pozz]

One other point: The way to fight is to make sure you can defend your technical points. Examples: @amirm's testing of:

• Mytek (over incorrect MQA implementation, no less)
• Schiit
• Denon

 

[pozz]

Then one day, Tidal re-jigged their tiers- lossy (as it is now), redbook (so proper "lossless") and Hi-res (everything above 16/44 or 16/48) BUT- to access Hi Res you had to buy a "Tidal Hi res" certified DAC.

Good point.

 

[dmac6419]

There are codecs that achieve this for a fraction of the filesize and aren't trying to squeeze money out of every step of audio production and consumption chain while using underhanded tactics to do so.

If you don't subscribe,how can they squeeze,last I checked we/you have choices, am I missing something?

 

[KSTR]

The triangle diagram itself is already misleading. It assumes the content is encoded as frequencies but it's not, it is time-domain data. A spectrum assumes you are looking at some significant time window of the sample stream (up to the whole stream/file size) and then it is true that a typical music spectrum looks like that, a pink spectrum, and that's what they are showing here...
But if you look short term spectra (which have reduced frequency resolution by definition), there is content with high frequency and high level at the same time which immediately violates the triange boundaries.

This is a (rather loudness-war ruined) section of a commercial CD, steeply highpass-filtered at 11kHz:

Baam! Hitting -9dBFS peaking at 13kHz, way outside of the safe triangle.
Therefore, a good test would be trying to encode some material which contains short isolated high frequency high energy "blips" buried in normal music... or just find any EDM track with this kind of signals.

 

[pozz]

If you don't subscribe,how can they squeeze,last I checked we/you have choices, am I missing something?

Try finding a commercial multichannel recording that doesn't somehow have to pay licensing fees to Dolby. Or deal with the hardware compability issues today without running into licensing again.

This is part of the reason for the dislike. The creation of paywalls, tiers and other boundaries that add to complexity and expense.

 

[thefsb]

Not saying MQA aren't a bit shifty, what company that is trying to sell something isn't.

Some really aren't. But even if you don't think MQA's behavior qualifies as deceptive or snakeoil, it's utterly bizarre that they defend their work in the way they do. MQA's responses to GoldenOne say in effect that their codec only works for certain kinds of "natural" music. HUH? Will they specify that in quantitative terms please?

 

[thefsb]

Thank you @GoldenOne. Great work and that's a tremendous write up. I cancelled my Tidal a few minutes ago after reading it.

ASR really is the business, eh? So much good info here helping us dimwits through the audiophile quagmire. Or, dimwits like me, I should probably say.

 

[LaL]

If MQA is eventually more advantageous in copyright protection
how much influence could the Record Labels have in regards to
what file type the music streamers are allowed to use.
It could possibly give them more control if they only distribute their music in MQA.

 

[thefsb]

Phillips and Sony did a decent job 40 years ago when they chose 44.1k 16-bit PCM for the Compact Disc.

I spent 25 years in telecom and data communications systems engineering and part of my work was standardization. So many standards. I guess you could arrange them in a quadrant analysis with technically good to bad horizontal and market impact vertical. 44.1/16PCM is in one of the corners.

 

[abdo123]

Possibly- but his continued engagement with the threads like this rebutting some of the challenges to MQA might give some cause to think he does care, just not in the direction they would like. Personally, his or anyone elses views on it doesnt bother me either way. I've seen enough to make my mind up about it - my hope is that is becomes more of an irrelevance rather than gains any further traction.

He cares that this site does not become a place where misinformation is propagated, which is why he is correcting people.

The fact that this topic is about a codecs and that he spent a big chunk of his career developing one himself might have also agitated him further than these things normally do.

 

[raistlin65]

MQA proposes another form of lossless that says it music follows statistical distribution, they are able to fully preserve its actual informational content including what is in ultrasonic. If they are able to achieve this fully, what is your beef with it? That it doesn't spit out a bunch of bits for a rectangular channel is not used in music?

The keyword here is "proposes."

Until it has been independently verified that their encoder achieves that, we should be very skeptical that it does. And so far, MQA appears very resistant to having its claims validated by others. Which should make us even more skeptical. Snake oil products often try to muddy the definition of terms in audio so that they can market their products, and then they refuse to verify their claims. We should be worried that might be happening here.

That being said, other than its geek value to people interested in the encoder as a technology, it has almost no practical value to consumers.

The fact that it can preserve some >20khz doesn't really matter. The very large majority of musicians don't hear over 20khz. Nor do most of the people doing the mixing and mastering. And even if the musicians and the audio producers can hear it, most will have no idea if the speakers or headphones they are using to verify their "artistic intent" are being accurately represented in their recordings. You aren't measuring the frequency response of headphones and speakers above 20khz. Who's doing that so that they know how their playback equipment is performing?

Meanwhile, most consumers don't hear above 20khz. And same issue with consumer's playback equipment. Where are people getting EQ calibration for +20khz for their headphones, assuming their headphones even do +20khz? Are people's calibration mics for REW calibrated for above 20khz? (And then there's the big question of whether or not above 20khz adds anything to the aesthetic experience for those who can hear it with their ears and their equipment, where the artist's intent also was to provide it?)

So assuming that what MQA says is true--and that's a big assumption--we have a technology that has an extremely niche use case of preserving audio that most consumers won't benefit from. And yet, it's being marketed as good for a broad base of consumers so that MQA can make a licensing fee. People that aren't even interested in MQA are paying the MQA tax when they buy a new DAC.

That's our "beef" with it, to use your words. And you should be able to acknowledge that. You seem to enjoy geeking out about how it theoretically works, and that's great. There's nothing wrong with that. But you can do that and not be dismissive of the lack of practical benefits it offers. It doesn't really matter if it works as they say it does.

 

[levimax]

If you don't subscribe,how can they squeeze,last I checked we/you have choices, am I missing something?

For now it appears we have choices but it has been documented that some MQA encoded files are being provided to the streaming services and served up without notifying the consumer. It has been suggested by @Amir that these limited cases may just be an "accident" but it is not hard to imagine that this is exactly what MQA is planning to do and it makes a lot of sense for the record labels to do so. Much of the opposition to MQA is that if they are successful with their business plan that choice will indeed be eliminated and we will all be forced into a inferior closed system. Good for MQA and the record companies but not for the consumer. @Amir does not believe this will happen but if you read @JohnYang1997 comments it is already happening.