MQA is Vaporware

[fung0]

49 minutes ago, PeterSt said:

There is no warmer sound in MQA. The contrary.

 

Very well, then: "the greener sound of MQA." All subjective judgments are ultimately alike. (Especially metaphorical ones.)

[PeterSt]

50 minutes ago, fung0 said:

Very well, then: "the greener sound of MQA." All subjective judgments are ultimately alike.

 

That depends a little on how concrete one can be in his judgment which should be a generally readable one. Like "more airy" or "more compressed". Or "more open" as a sub form of airy.

My point : when you are serious with this, it is not subjective. Nothing in the end is; I have been auditioning systems (also our own) with quite many people and one thing is easy to notice : you always agree without many words. But, the people I'm talking about are, say, experienced. With less experience it also works but then it needs pointing out first. Right after that the phenomenon (which ever it is) is recognized by the "student". Matter of complication does not matter.

 

MQA in general sounds more white.

This should mean :

 

Less color in the cymbals (white sound is profound and not "copper" sound);

More robotic playback;

And the flavor itself (which is never a good thing).

 

Of course there is more to say about MQA, but for now I was only responding to the more warm of which I think the opposite is true.

And yes, my judgment can be personal and for that reason subjective, but *only* because my equipment is personal compared to yours. It is just different. And this does a few things.

 

 

[FredericV]

1 hour ago, PeterSt said:

MQA in general sounds more white.

This should mean :

 

Less color in the cymbals (white sound is profound and not "copper" sound);

More robotic playback;

And the flavor itself (which is never a good thing).

Something in female voices was lacking for me in the MQA version vs the DXD version. There was the illusion of more echo & reverb, but at the cost of everything being thinner. When archimago posted the impulse responses of the renderer, it made sense why it sounded that way.

I yet need to find a piece with a lot of cymbals where I also have the original file that was used for the encode. Do you have such fileset?

I did this test on a Brooklyn + Vitus RI-100 + Amphion Two 18 in nearfield. In the same setup, I now also have a Manhattan 2 which I'm burning in for a few days. Will the Manhattan enlarge these differences? I intend to find out.

I have the Manhattan mk2 because it is a great DAC independent of MQA, and it's also a DAC that is used by the studio's.
If the Manhattan mk1 did not have the need for a custom linux driver, I would probably have bought the mk1 already. We used it on a show last year and the feedback was very positive.

I actually helped the Linux driver to recognize the Manhattan mk1 by tracing the correct USB ID's, and Jurgen patched it in:
https://github.com/lintweaker/mytekusb2/blob/master/chip.c
 

The manhattan mk2 just works out of the box on linux just like the Brooklyn.

[PeterSt]

1 minute ago, FredericV said:

Something in female voices was lacking for me in the MQA version vs the DXD version. There was the illusion of more echo & reverb, but at the cost of everything being thinner.

 

Frederic, FYI and FWIW :

I see no legality in using anything that for sure utilized the frequency above, say, 48KHz, under which we may regard the DXD of 2nl.

We can of course deliberately try to debunk MQA while knowing it won't have any higher "native" (blurp) resolution than 48KHz (96KHz sampling rate), but I don't like to be part of that league. So for me counts : a 96KHz sampled file is still theoretically OK for MQA, but a 352.8KHz (DXD) is not (because it will be upsampled from 88.2 to 352.8). And thus, for me, there is no reason to even start thinking of comparing DXD with MQA (MQA will always lose for a for me too obvious reason).

Otherwise it is my idea that MQA is not thinner at all. For me it is a kind of opposite : it is clearly more firm. I can hear it in the upper bass (say like a Level 42 slapped bass) which in my view also expresses in the mike-in-mouth ("look into the mouth") voices which you see report everywhere. I heartedly agree with that. This too is the opposite of thin. And before people misunderstand : this is not related to the "bass" phenomenon at all. If anything, "bass" as such receives more toutness and the net result of that is less bassy sound (frequency will really be higher).

 

14 minutes ago, FredericV said:

I yet need to find a piece with a lot of cymbals where I also have the original file that was used for the encode. Do you have such fileset?

 

No. And this is already because the masters used for MQA are different (forget about 2nl for now). And otherwise I wouldn't be able to tell, I guess.

 

 

 

[mansr]

6 minutes ago, PeterSt said:

I see no legality in using anything that for sure utilized the frequency above, say, 48KHz, under which we may regard the DXD of 2nl.

Well, the MQA marketing claims it can accurately convey a 352 kHz master. They even go out of their way to include the original sample rate in the metadata so DACs can display it in place of the true 88/96 kHz.

 

Imagine someone selling aspirin with the claim that it cures headaches and cancer. What you're doing is equivalent to saying, "oh, it obviously can't cure cancer, it's only aspirin, but it does seem to work against headaches, so it's all good."

[JoeWhip]

In regards to the analogy with 4K video, while the increase in resolution is not all that significant, the use of HDR is. It is something that greatly improves the viewing experience, when used properly. It us a real advance in the presentation if video. I watched Wonder Woman in 4K the other day in an OLED and it was mikes better than the SDR version because of the use of HDR. So the actual benefits of 4K are obvious even from across the room. MQA, not so much.

[Jud]

8 hours ago, crenca said:

prices have and continue to rise at "luxury" rates of increase

 

I bought my DAC for $375.  The previous DACs that I purchased were ~$2000 (more than 25 years ago) and ~$500 (about 7 or 8 years ago).  I like this DAC better than either of those.

 

The prices on speakers, however, have increased, though I don't know what the comparison to the rate of inflation is.  (The speakers I purchased 30 years ago, the Vandersteen 2 series, are about double the price today.)

 

So I think the story is considerably more complex.

[FredericV]

25 minutes ago, PeterSt said:

So for me counts : a 96KHz sampled file is still theoretically OK for MQA, but a 352.8KHz (DXD) is not (because it will be upsampled from 88.2 to 352.8). And thus, for me, there is no reason to even start thinking of comparing DXD with MQA (MQA will always lose for a for me too obvious reason)

 

That's why the first unfold does not have these issues. It's a band limited version of the master, without the new issues introduced by the second unfold. If MQA was just clever compression/folding without the issues plagued by the second unfold, and not calling this authenticated, I don't think we would even have this thread.

The first unfold is the clever part of MQA, usable for a limited time in a world with limited bandwidth in some parts. This is what Bob explains in the video from Jaap Veenstra, when he very cleverly works around the question why MQA has no +48Khz content (so equivalent of 24/96 sampling in terms of bandwidth). He discusses traffic caps and very expensive mobile subscriptions, which cost a multiple of the Tidal subscription in some countries. He has a point.

These advantages are only temporary. There will be a time when qobuz streaming will not be expensive in 24/96 flac via mobile in these countries. In Belgium, my 1 GB mobile limit has increased to 12 GB in just a few years. So what if it becomes 100 GB in a few years. That's a lot of flac mobile streaming. With 5G being introduced, these caps will automatically be incremented by not a small fraction,.
 

[firedog]

5 hours ago, PeterV said:

I wonder what is your opinion Chris Conaker and all other readers regarding the latest post of Frederic vanden Poel who does not understand what privacy really means. Sending an e-mail to my employer in a pathetic attempt to smuther my opinion regarding this new format. Then the witchhunt began, incriminating me that I am being paid by MQA. Total paranoia acting by him, checking my Facebook account where I work..and if something has been changed on FB,  he accuses me for being shill and paid by MQA.. ! So that is the reason why I am now fighting back and acuse him to be a fraud who is convinced that all music must be played at 432 Hz. This occult conviction is the real reason why his hate towards MQA is over the top and his motivation to use every means to 'kill' it. All my posts on CA and Archimago and others over the last 2 years by ne are of private nature and not related to my employer or any other imaginary employer which would be MQA in the mind of this sick conspiracy theory follower who believes all music is being recorded distorted anyway at the wrong frequency.  This has to be stopped and I request you Chris to ban us both now. Have fun everyone with this stupid format war. I am done, will just proceed loving my hobby listening to music in either PCM or PCM+ as I would define MQA. 

 

Frederic: be happy with your hobbies and convictions. I warned you not to interfere and that we would not fight and reply to each others posts. You could not resist, but now you have severly crossed a line. 

My opinion, as previously stated, is that we are tired of hearing about it. Take it off public forum or start your own thread about it. 

[PeterSt]

52 minutes ago, mansr said:

Well, the MQA marketing claims it can accurately convey a 352 kHz master. They even go out of their way to include the original sample rate in the metadata so DACs can display it in place of the true 88/96 kHz.

 

Oh yes - which is the first problem you run into with a software player. So I can tell you and everybody from experience that this behavior is super-sneaky. This is how it goes :

 

1. Get the native sampling rate from the metadata;

2. Show this is the player;

3. Play it and via indirect fashion show the bit rate as the sound engine discovers it.

 

For a 192KHz denoted file this resulted in my case in the showing of the 192KHz everywhere and I couldn't understand what I saw. The Doors in native 192KHz ?

It was not true.

 

I can not say that this happened all on MQA ltd's purpose, but for sure in my case it went that way, just because I followed directions (which seemed logical to me).

Now hardware :

The very same should happen, without actual possibility to check what the DAC etc. would do for real. It is a DIS-PLAY.

 

Maybe some conspiracy thinking brings new ideas regarding this. In any event, I now watch the 96 and 192 logos and feel I am influenced by it. Sampling rate is always the same, but the logo tells me different.

 

Quote

Well, the MQA marketing claims it can accurately convey a 352 kHz master.

 

I know. What I tried to bring across is that while we now know how reality is (at least we seem to think we know), it is unfair or self-cheating if we use 352.8 as an argument for more poor sound in MQA. It just *is* more poor all the way. So I don't look at it (not even for testing the 352 logo, LOL).

[mansr]

It's graph time again. Today we're looking at 2L-109 from the free samples.

 

This is the full spectrum of the DXD original. There the high frequencies have mostly noise with a few narrow spikes. Zooming in on the one at 66 kHz:

The spike extends about 20 dB above the surrounding noise floor, the tallest part about 200 Hz wide just below 66 kHz.

 

We know MQA downsamples this signal to 88.2 kHz as part of the encoding process. If the filters used here allow aliasing, this spike would land at 22.2 kHz and extend upwards about 200 Hz (the alias frequencies are mirrored). Looking at the spectra of the original and the decoded MQA (before "rendering") around that frequency, this is what we find:

There is indeed a difference between the original and MQA exactly where expected. It even has the same three distinct peaks. The difference is only about 0.3 dB, but it is there nonetheless.

 

From this we can tell that the attenuation of the anti-aliasing filter at 66 kHz is roughly 20 dB. We still know nothing about the shape of the filter. This could be the final stopband attenuation, or it could partway down a slope of unknown steepness. Either way, it is quite poor. A good filter would have at least 120 dB attenuation at this point.

[firedog]

6 hours ago, PeterSt said:

 

What if this is related to experience ?

I Read your post, firedog, and suddenly it came to me that there is a common denominator between this what you call press (but which is often reviewers and mastering engineers) and me : day in day out listening very explicitly for improvement. In my case of my own gear, but there is not much difference with "different gear" if you see what we achieve with software upgrades, hardware upgrades and USB cables for that matter. I am sure I would be as  enthusiastic as the press you refer to, just because it is the most easy (for me, by now) to find these great MQA albums. The point is that it seems to be so that this does not count for each and everything, plus I hear a flavor in the sound (which in this case is not killing).

 

In the context of the Boris Blank album PeterV made the remark similar to "great that MQA even works for electronic music". WRONG ! it is MADE for that. I already noticed with Yello - Toy (which is 48KHz IIRC) that it sounds way better (or more emphasizing the goodness of MQA) than other albums. This Boris Blank no different. And at least that one I never heard of before, so I can ben quite objective.

 

All it would need is that MQA showcases pick the albums which work best for MQA and show that to the press et all.

So what I am saying is that it would be very easy to let sound MQA way better than normal Redbook and certainly (all this failed) Hires.

And mind you, I am working on this since April or so. And contrary to most of us here, I want it to work.

Peter-

If I understood your point, I'm not sure it is a correct supposition. 

I'm a fairly critical listener, especially when I decide to A/B test something like MQA vs. non-MQA. I'm more than willing to admit that I'm not the world's most discerning listener, and not as good as some professionals, but certainly way better than the average listener.

 

My experience with MQA is mixed: some albums sound better to me; some not too different, some worse. In none of the cases are any of the differences something I would term "revolutionary," "ear-opening", "completely different", or anything approaching the descriptions I've read by many people.  They sound to me something like the same level of I hear when comparing different masterings of the same record - I can hear a certain difference, and sometimes I think it is an improvement.

 

I'm pretty sure that if MQA was making the revolutionary and game changing difference I keep reading about, then it should be obvious to almost anyone, much like the difference between old standard TV and HD TV is (even when it is compressed), or as Chris said, the improvement of 1080p over the old standard - you see it right away and it is clearly better. 

If even the small differences were consistently positive for MQA, I'd have a better opinion of it, but that isn't what I hear. I don't see any great advantage to it. 

 

And again, if it was just another format choice, I'd be fine with it. I could choose to get an MQA remaster or not in many cases. But I don't think that is what is going to happen and I don't think that is the intent. 

 

I think that if the format gets well entrenched, we will see the use of it's DRM features to restrict access to better SQ, to eliminate non MQA versions, and (especially hi-res ones) and  to promote discriminatory pricing schemes. Why? Because there isn't another way to make money from it.  

 

 It's already clear that there isn't some great mass market for hi-res and even CD quality streaming. If there was, we already would have seen some massive migration from Spotify, Google Music, and Apple Music mp3 to the Tidal Hi-Fi tier. But that didn't happen, and I haven't heard of anyone who thinks it will.

 

The long term goal of the MQA format is almost certainly not simply to provide, at no extra cost, another flavor/master of each album available for streaming. That is a money loser for all the parties involved. It's being provided that way as a loss leader today. An investment and test of the market before future developments - developments which will cost money and probably restrict choice in order to ensure that people pay for the use of MQA. 

[PeterSt]

1 hour ago, FredericV said:

He discusses traffic caps and very expensive mobile subscriptions, which cost a multiple of the Tidal subscription in some countries. He has a point.

These advantages are only temporary.

 

Yes. But here too, why put emphasis on it while "we" already seem to know the rubbish in that ?

So for us it can be about one thing only : better SQ. And with the danger of running in circles, that can only happen when the native file (say the one we stream from Tidal) had "better data" in it than all the other Redbook. Stupid thing is (sadly ? luckily ?) that the masters are different and as far as I can see almost exclusively better (less compression). And that creates the sound too. But it should be so that the deblurring happened in there. Whatever that means for real and to what degree it ever can be for real.

 

Another kind of conspiracy thought could be : why on earth ARE those masters all different ? they are you know.

If you ask me they all did not exist (that I can tell - again please forget 2nl). So envision a process that apparently can run in/on each production (???) facility that in retrospect is able to recollect some kind of native ever used for production master file with less compression in it than the one which went out, and apply whatever MQA on to it.

Can that exist ? I don't think so.

Are the mixes different then ? sure not (rare cases let alone).

 

So how odd would the chances be that there's just this ever used for production master file and that some decompression was applied to it, and which would MQA ltd call "MQA" (it will be a unique process, also depending on the compressors and limiters etc. used).

 

Yes, I think I just had a good idea with this. It would make many things suddenly plausible.

 

Listen again to the stuffed ears machine head album. The other day I listened to it I had the idea of it sounding so smooth because all was made flat (so, large compression, without that being visible really because of the type of music - as I now think about this twice). Now listen to the MQA version of it. Cymbals are louder than on whatever version of it I could discover. How ?? Decompress it in a smart way and pay attention to the higher frequencies (let those be). What you'd get is a more open sound.

Try it.

[realhifi]

43 minutes ago, firedog said:

My opinion, as previously stated, is that we are tired of hearing about it. Take it off public forum or start your own thread about it. 

I disagree. This exchange has actually been enlightening. Not sure but did PeterV get banned or did he just decide to leave?  Also, is FredericV a manufacturer of docs?

[Fokus]

23 minutes ago, mansr said:

We know MQA downsamples this signal to 88.2 kHz as part of the encoding process. If the filters used here allow aliasing, this spike would land at 22.2 kHz ...  this is what we find:

 

There is indeed a difference between the original and MQA exactly where expected.... The difference is only about 0.3 dB, but it is there nonetheless.

...  Either way, it is quite poor. A good filter would have at least 120 dB attenuation at this point.

 

MQA have never hidden that their downsampling strategy selects the leakiest-possible filter that puts the actual aliasing artefacts at roughly the same level as the original signal's <20kHz noise floor.

 

So this is not new.

 

 

[FredericV]

37 minutes ago, firedog said:

I think that if the format gets well entrenched, we will see the use of it's DRM features to restrict access to better SQ, to eliminate non MQA versions, and (especially hi-res ones) and  to promote discriminatory pricing schemes. Why? Because theire isn't another way to make money from it.

 

This my biggest concern with MQA. For now the sox method works great with MQA files as an unofficial alternative. But if they start to enable crippling features it probably won't. The main problem is that we don't know if these features which are in the patent, are actually in MQA:

This is the patent that describes features to cripple undecoded versions:
https://www.google.com/patents/US20160005411

Now Bob Stuart claims in the interview with Jaap Veenstra that not all their patent applications apply to MQA. At the same time he denies there's crypto in MQA. But why does MQA uses crypto?

https://www.infosecurityeurope.com/__novadocuments/266589?v=636078256117430000

This is the interview where Bob denies the crypto:

What is also interesting is in the first minutes he debunks the need for 32 bit and also has some reservations about 24 bit.

At the same time, they are promoting a new 32bit fixed/192dB/352.8k DXD ADC and complete mastering workflow at their latest AES convention, while MQA encoding throws most of that away. In the video from Jaap, we actually learn that MQA is somewhere around 17 bits. So why capture in 32 bits 352.8K to keep an actual resolution of 17/88.2 (as DXD = 8x 44.1, they use 88.2 as first unfold for DXD) ?

http://www.aes.org/events/143/recording/?ID=5693

 

 

[FredericV]

7 minutes ago, realhifi said:

I disagree. This exchange has actually been enlightening. Not sure but did PeterV get banned or did he just decide to leave?  Also, is FredericV a manufacturer of docs?

 

What are docs? My affiliation is in my signature. But let's not change the discussion into my product. This is the MQA topic and I would like it to stay that way.

[realhifi]

11 hours ago, The Computer Audiophile said:

I don't believe 4k video was something consumers asked for or need. I'm not sure it's actually that good for anyone other than those with giant screens who sit close, and the people selling goods. 

 

Lossless 1080p, which isn't sent into homes via streaming, satellite, or cable, would be better. When people see how good an over the air HD video looks compared to lossy 4k, they may think twice about 4k. The same can be said for great 44.1 versus high resolution and MQA.

I’m assuming you’ve spent quite a bit of time in front of a 4K projector and screen seeing the differences between 1080 and 4K?  Not to mention HDR which is a bear to transmit over HDMI cables because of the intense amount of information it is passing. Not sure I’d equate the difference between 4K and 1080 as the same as 44.1 and 24/192 which I can have problems discerning between. 

[realhifi]

Just now, FredericV said:

 

What are docs? My affiliation is in my signature. But let's not change the discussion into my product. This is the MQA topic and I would like it to stay that way.

Oops , meant dacs. Sorry about confusion. 

[realhifi]

1 minute ago, FredericV said:

 

What are docs? My affiliation is in my signature. But let's not change the discussion into my product. This is the MQA topic and I would like it to stay that way.

I don’t believe I was trying to change the discussion at all but merely trying to understand the two people involved. 

[mansr]

8 minutes ago, Fokus said:

MQA have never hidden that their downsampling strategy selects the leakiest-possible filter that puts the actual aliasing artefacts at roughly the same level as the original signal's <20kHz noise floor.

 

So this is not new.

I haven't seen it verified with actual files before.

[Fokus]

True. You need to get lucky with a file with low noise and an easily-identifiable ultrasonic artefact.

[psjug]

1 hour ago, FredericV said:

The first unfold is the clever part of MQA, usable for a limited time in a world with limited bandwidth in some parts. 

 

MQA may be clever, but in a Rube Goldberg sort of way.  As Miska pointed out a long time ago, why not just use 18/96 Flac?

 

[FredericV]

1 minute ago, psjug said:

 

MQA may be clever, but in a Rube Goldberg sort of way.  As Miska pointed out a long time ago, why not just use 18/96 Flac?

 

 

There are some blanking initiatives, where the noise bits of the 24 bit capture are zeroed. As noise can't be compressed well when it needs to stay lossless, it costs a lot of bandwidth. Blanking the noise can achieve compression factors which are as good as MQA. It does not waste entropy thus bitrate can be reduced.

I believe both Dr Lesurf and Xivero suggested such workarounds.

http://www.audiomisc.co.uk/MQA/cool/bitfreezing.html
https://www.xivero.com/xifeo/

[The Computer Audiophile]

15 minutes ago, realhifi said:

I’m assuming you’ve spent quite a bit of time in front of a 4K projector and screen seeing the differences between 1080 and 4K?  Not to mention HDR which is a bear to transmit over HDMI cables because of the intense amount of information it is passing. Not sure I’d equate the difference between 4K and 1080 as the same as 44.1 and 24/192 which I can have problems discerning between. 

I agree the differences in video are much easier to see, but if 1080p was delivered better, 99 percent of people wouldn't care about 4k.