Jump to content
IGNORED

MQA is Vaporware


Recommended Posts

1 minute ago, manisandher said:

 

Yep, it was 1.

 

FWIW, to my ears, the MQA does sound different to the hires... but not necessarily better.

 

Nowadays I'm more than happy with simple redbook. Upsampling in XXHighEnd or HQPlayer and sending to a decent (preferably filterless) DAC, redbook can sound truly stunning.

 

Mani.

Hi Mani,

Thanks. Do you have the time domain difference as the difference expressed using dB's ? You have presented the spectrum, but not the time domain difference. Thanks.

 

Regards,

Shadders.

Link to comment
20 minutes ago, charlesphoto said:

As for downloads, Qobuz hangs up on me all the time although I have fiber-optic speed.

 

I have Lee on ignore, so the above isn't Charles, but him quoting Lee.

 

Lee, then quite plainly something else other than your Internet speed is wrong, so this really isn't relevant to the discussion.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
7 minutes ago, pkane2001 said:

But of course :)

 

4 minutes ago, Shadders said:

 

  1. Has anyone used this software to compute the difference of regular and MQA files in the time domain ?

 

Happy to give it a go when I have some time.

 

Mani.

Main: SOtM sMS-200 -> Okto dac8PRO -> 6x Neurochrome 286 mono amps -> Tune Audio Anima horns + 2x Rotel RB-1590 amps -> 4 subs

Home Office: SOtM sMS-200 -> MOTU UltraLite-mk5 -> 6x Neurochrome 286 mono amps -> Impulse H2 speakers

Vinyl: Technics SP10 / London (Decca) Reference -> Trafomatic Luna -> RME ADI-2 Pro

Link to comment
4 minutes ago, Shadders said:

Do you have the time domain difference as the difference expressed using dB's ? You have presented the spectrum, but not the time domain difference.

 

As Miska pointed out earlier, it's actually quite large in the time domain:

 

On 3/12/2019 at 2:18 PM, Miska said:

 

Thanks!

 

Signal Max  : 2.66631e+06 (-9.96 dB)

 

image.thumb.png.c8a9289c114563da0376d4ec45439cb2.png

 

Pretty high level. Not sure how accurate the diff is...

 

image.thumb.png.cc2a92b9a8fdc96a9e7ca080bea8b976.png

 

 

Again, this may be down to lack of perfect alignment.

 

Mani.

Main: SOtM sMS-200 -> Okto dac8PRO -> 6x Neurochrome 286 mono amps -> Tune Audio Anima horns + 2x Rotel RB-1590 amps -> 4 subs

Home Office: SOtM sMS-200 -> MOTU UltraLite-mk5 -> 6x Neurochrome 286 mono amps -> Impulse H2 speakers

Vinyl: Technics SP10 / London (Decca) Reference -> Trafomatic Luna -> RME ADI-2 Pro

Link to comment
1 hour ago, Lee Scoggins said:

 

This is an analogy that is too simple and misleading.  You cannot look at this solely on an individual consumer's experience and make the claim when the whole thing is about distributing music at scale.

 

Running a streaming service requires lots of bandwidth and lots of storage.  While individually both are cheap, there are many issues that these services face at the vast scales they work in.  Look at amazon or netflix.  Many times the video will stutter at night when bandwidth is heaviest. As for downloads, Qobuz hangs up on me all the time although I have fiber-optic speed.  24/192 I think requires around 9 mbps download speeds.  It's provable that many areas in the country don't have that.

 

As long as these problems persist, you cannot argue that file size compression is an idea whose time "has come and gone."

 

Umm, with respect, this is not how it works any longer. Apple has pretty vast resources, but the newer streaming services? 

 

Nope.

 

They use Amazon or IBM or other similar services - essentially they  pay for what they use, and nothing more. And even that will be at a very steeply discounted rate.  

 

It is a very effective economic model. It also means that compression is much more effective at the storage level, since that is the constant cost. And that compression absolutely must be lossless. The reasons that is true are fairly obvious. In point of fact, most transmissions are already compressed, being decompressed at the terminal end. 

 

Without getting into a lot of technical verbiage, I am afraid to convince anyone that MQA compression is valuable, technical verbiage is exactly what is needed. Okay, by that I mean more detail on how they are storing information and then retrieving it. Mmm...

 

Ask them to explain more how they have convolved the data and are avoiding/handling the traditional issues involved with a homomorphic system when they deconvolve it to retrieve the data. The point is, there are a *lot* of problems in even a simple system to do that, and it explains why their filters are shaped the way they are. It also explains why they may have came up with something that works in a novel or useful manner. They will have to explain that though. 

 

Perhaps it would be useful to be more specific here, if in a very basic sort of way. Traditionally and conventionally, they would use a homomorphic transform to convolve the audio signal with an impulse response of a delta func and a shifted and scaled delta func. That would normally be a Fourier transform followed by a logarithm.  Then linear filtering is or can be used to separate the data back and reverse the homomorphic transform. This is nothing new nor proprietary, though implementations of it can be. You can probably find it in any DSP textbook or reference. 

 

What is interesting in this case (though I am not claiming it to be the situation with MQA) is that in the linear filter, the frequency domain is processed the way one would normally process the time domain. In a FFT, for instance, the spectra being multiplied would be the time domain. “Deblurring” anyone? 

 

Anyway, I horribly simplified that, and I am sure others will probably correct it, but I do think some discussion on the technical level is needed from MQA. They are plenty smart enough to explain what they are doing in a way to pass the Feynman test and yet not give away their technical secrets, whatever those are. None of this stuff is anything a sophomore would blink an eye at.  They should talk about it at that level. 

 

Hell, if I can think of it and see the parallels with very common DSP, it is a sure bet other people already have, probably thought deeper and with more rigor.  ;)   The marketing speak about this kind of stuff not withstanding, only solid information is going to save their case.

 

This is also why even a couple years ago, people (myself included) expressed concern about DRM.  Surely you can see that the technical MQA secrets will be exposed or duplicated without too much delay. They are almost certainly based upon very well known  prior art. Once people know something has been done, it is inevitable other people will duplicate it. History is replete with examples of near simultaneous inventions. That is true even if my suppositions went down a wrong path. I expect my thinking is pretty close to the facts here, but it is pretty conventional thinking. 

 

Okay, off my soapbox now. 

 

MQA should be working hard to engage with people like Chris to help disseminate accurate knowledge and advance their cause, rather than pushing them away with the effing bad behavior and bullshit they demonstrate now.

 

Man, they should be ashamed. 

 

- Paul

 

I looked for typos, but they will be in there. This stupid iPad insists on correcting words like convolve to something else for example. 

Anyone who considers protocol unimportant has never dealt with a cat DAC.

Robert A. Heinlein

Link to comment
1 minute ago, manisandher said:

 

As Miska pointed out earlier, it's actually quite large in the time domain:

 

 

Again, this may be down to lack of perfect alignment.

 

Mani.

Hi,

Thanks. Was the difference based on the MQA analogue output signal ?

 

If so, then the blurring of the signal by the MQA filters will account for that. - how significant ????

 

Would need to be determined by calculating the difference of the MQA signal before the MQA filters and the non MQA signal.

 

Has anyone done this ?

 

Regards,

Shadders.

Link to comment
1 hour ago, John Dyson said:

Okay -- I do agree that data compression of audio is of less importance than today.  However, there are still cases where compression might be useful.  Maybe a LOT of compression of good quality might be more useful than some compression at the highest quality.  Where bandwidth reduction is needed -- then a lot of reduction is needed.  Data reduction for storage reasons is of less importance (or the need is less common.)

When I need data storage size reduction, I am happy with flac -- it does just enough to be worthwhile, and except for material that has data out of range, is good enough to store the data that I normally use floating point for manipulating.  Usually, flac is otherwise good enough.

I have to admit, sometimes I do play with flac's apodization options to maximize compression -- but the benefit is similar to that of twiddling thumbs.

 

MQA is optimally unoptimum in yet another way -- time has come and gone -- after digesting the informaion/thinking about it --  I do agree.

 

John

 

John, something else I was wondering about regarding your remark that MQA's lossy compression is applied to an area where it makes the least sense from a technical point of view:

 

Because there really is not a need for smaller file sizes in the first place, and even if there is, non-proprietary lossless compression will work just fine, I conclude with pretty fair certainty that the lossy compression must be to protect IP. 

 

(I'm a lawyer who actually did copyright litigation, back when dinosaurs roamed the Earth.  If a client came to me today in MQA's position and asked how to protect their IP, I would say patent isn't sufficient; you'd want trade secret and copyright as well, and to help with the trade secret and really put some DMCA teeth into the copyright protection, you'd want an anti-copying measure that people would have to work at bypassing to get to the IP.  Lossy compression with a crypto key to decode is ideal for these purposes.)

 

Do you suppose the reason the lossy compression is primarily used in a range where it's technically sub-optimum is because this is the range where MQA's supposed "deblurring" IP should operate?

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
3 minutes ago, Shadders said:

Hi,

Thanks. Was the difference based on the MQA analogue output signal ?

 

If so, then the blurring of the signal by the MQA filters will account for that. - how significant ????

 

Would need to be determined by calculating the difference of the MQA signal before the MQA filters and the non MQA signal.

 

Has anyone done this ?

 

Regards,

Shadders.

 

No, everything remained in the digital domain. You can learn more about it here:

 

Mani.

Main: SOtM sMS-200 -> Okto dac8PRO -> 6x Neurochrome 286 mono amps -> Tune Audio Anima horns + 2x Rotel RB-1590 amps -> 4 subs

Home Office: SOtM sMS-200 -> MOTU UltraLite-mk5 -> 6x Neurochrome 286 mono amps -> Impulse H2 speakers

Vinyl: Technics SP10 / London (Decca) Reference -> Trafomatic Luna -> RME ADI-2 Pro

Link to comment
13 minutes ago, The Computer Audiophile said:

You're losing the plot. 

 

Your version of this would have been abrasive and turned people off 😁

 

On the contrary, I saw the plot from the almost very beginning.  Your the one who took convincing.  

 

The "abrasive" turn off is not calling a spade a spade, but endless technical and non-technical "debate" with astroturfers and trade publication writers who have no intention of acknowledging the truth or anything else but their own interests.  

Hey MQA, if it is not all $voodoo$, show us the math!

Link to comment
2 minutes ago, manisandher said:

 

No, everything remained in the digital domain. You can learn more about it here:

 

Mani.

Hi,

Thanks - 12 pages.

Does anyone know if the two digital files - MQA and High Resolution, were compared with the MQA signal BEFORE the MQA filters were applied ?

 

The MQA filters, despite their low tap number, are non-linear, so may have significant impact upon the signal.

 

That is, the claimed deblur is near non-existent, and the actual differences are due to the MQA filters.

 

MQA Ltd have never, as far as i am aware, presented any evidence of the blur, such as a file difference example in the time domain.

 

Since MQA use blur to mean ringing and dispersion, they confuse the subject.

 

If there is ringing - where is the evidence - if it is so prevalent - where are the MQA Ltd examples of it occurring ??? They could easily point to every file processed showing the FFT with the ringing evident ???

 

Regards,

Shadders.

Link to comment
3 minutes ago, Shadders said:

Does anyone know if the two digital files - MQA and High Resolution, were compared with the MQA signal BEFORE the MQA filters were applied ?

 

Could you help me better understand what you mean here?

 

What I'm calling the 'MQA capture' is a 24/48 MQA file which has been decoded to 24/96, using Roon's MQA decoder. No MQA rendering has taken place. I've never quite understood whether any so-called 'MQA deblurring' is taking place in the decoder, the renderer, or both.

 

Mani.

Main: SOtM sMS-200 -> Okto dac8PRO -> 6x Neurochrome 286 mono amps -> Tune Audio Anima horns + 2x Rotel RB-1590 amps -> 4 subs

Home Office: SOtM sMS-200 -> MOTU UltraLite-mk5 -> 6x Neurochrome 286 mono amps -> Impulse H2 speakers

Vinyl: Technics SP10 / London (Decca) Reference -> Trafomatic Luna -> RME ADI-2 Pro

Link to comment
2 minutes ago, Shadders said:

No one is continually and persistently that thick to keep on repeating duff information.

 

On the contrary, I have seen LS do exactly the same regarding multiple topics.  I don't think it's anything unique to MQA, I just think it's a hyper-developed desire never to acknowledge when you might've been wrong.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
16 minutes ago, manisandher said:

 

Could you help me better understand what you mean here?

 

What I'm calling the 'MQA capture' is a 24/48 MQA file which has been decoded to 24/96, using Roon's MQA decoder. No MQA rendering has taken place. I've never quite understood whether any so-called 'MQA deblurring' is taking place in the decoder, the renderer, or both.

 

Mani.

Hi,

I could be wrong, but i thought the MQA encoded file was already deblurred - and once decoded, (the deblurred audio) is then passed through one of the 16 MQA filters (which are non linear) - and onto the DAC.

 

If it is possible to tap into the signal before the MQA filters - then we could see the difference between the MQA deblurred file and a normal high resolution file of the same song.

 

Regards,

Shadders.

Link to comment
3 minutes ago, Jud said:

 

I will (with civility 😜) disagree.

 

Writing like you're hollering at folks (or banging on desks?) is something people on both sides of an issue can often do equally well, and even when they don't do it equally well (for example, some people may just not like to do it), it's not always the technically correct side that has the advantage.

 

On the other hand, when you state simple technical truths in an understandable fashion, everybody very quickly knows which side has "no intention of acknowledging the truth," as you say, and we've seen that become very evident here.

 

As an analysis of everyday human communication and rhetorical strategery this is not true.  You were trained as a lawyer in an adversarial system - not in a scientific or 'truth' system where the facts are laid bear and "BAM!", everybody sees them and agrees.  There are very few human domains, if any, where "when you state simple technical truths in an understandable fashion..."  the good you say follows, actually follows.  If what you say were true then it would there would be world peace, the end of hunger, and unicorns and rainbows for every child.  

 

That ain't reality...

Hey MQA, if it is not all $voodoo$, show us the math!

Link to comment
41 minutes ago, manisandher said:

 

 

Happy to give it a go when I have some time.

 

Mani.

 

Here's a report. It's a large HTML file in a zip. Extract the HTML and open it in Chrome browser (or something other than Internet Explorer):

 

https://drive.google.com/open?id=1nB-pfdGINYYcR4ir8Yb2JU7Px_8s_qJ4

 

Regards,

 

     -Paul

Link to comment
2 minutes ago, manisandher said:

What I'm calling the 'MQA capture' is a 24/48 MQA file which has been decoded to 24/96, using Roon's MQA decoder. No MQA rendering has taken place. I've never quite understood whether any so-called 'MQA deblurring' is taking place in the decoder, the renderer, or both.

 

It seems to be sort of both... There is some nasty messy pre-filtering as part of the folding process. And then the upsampling filters, but those just leak images since they work from 88.2/96k upwards. As result they seem to use entire 22.05 - 44.1 kHz or 24 - 48 kHz band for rolling off the filter. I would call that blur, because it attenuates or removes high frequency components that make a transient fast...

 

For fast transient you need both short time and high frequencies which are of course mathematically directly related!

 

If you take for example most 176.4/192k hires recordings, in most cases there's nothing to deblur there because filter never modified the actual signal! However, now that they apply a filter to such hires that begins transitioning already above 22.05/24k frequencies they are actually trying to "fix" something that doesn't need fixing and also slowing down the intact transients...

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
15 minutes ago, charlesphoto said:

To me MQA has always seemed like a solution looking for a problem, as a way to dominate and potentially monopolize the source material for a niche hifi market. I know many people who could buy a complete dCS system at the drop of a hat but are perfectly content with the music channels through their TV, and maybe step up to a Sonos system and find it the end all be all. I know even in my own house, when my wife or kids want to start an after dinner dance party, it’s not about navigating the vageries of Roon and my boutique streaming set up to find the perfect resolution and master of an album, but “Hey Alexa! Play that “I Don’t Care” song” to the Play1 on top of the microwave in the kitchen. Sounds execrable to me, but everybody has a good time which is what most people have always wanted from their music. 

 

So in the big picture, hi rez and even cd quality streaming is a still a very small market and will probably always remain so, with not much pie to cut up. How many times have you mentioned Tidal to someone at a party and drawn a blank stare? Even I’m on the fence with my Qobuzz subscription whether I really need the hi-rez sub or not as I don’t find that much difference. 

 

What I find most abhorrent is there’s no need to give part of this already small pie to a bunch of software sham-sters when the artists themselves already receive so little. People who stream hi-rez do so full well with the knowledge they are doing so (and paying for it), and  typically don’t need it compressed or dumbed down, only to have to have special hardware to unpack it. It’s a rights grab to make money off the backs of others pure and simple, which is why we should all be outraged about it. 

 

This is the issue that MQA is trying to solve which is, "how do we get better sound if only a very small number are interested in high resolution digital?"

 

Agree or not, MQA's answer is to create a distribution approach that piggybacks it on popular streaming services.  If you give the customer easier access to hirez, you may get more fans of it.  And in the process you provide a way for the studios and labels to offer hirez and get more dollars for the artist by offering a premium tier. 

Link to comment
27 minutes ago, crenca said:

 

On the contrary, I saw the plot from the almost very beginning.  Your the one who took convincing.  

 

The "abrasive" turn off is not calling a spade a spade, but endless technical and non-technical "debate" with astroturfers and trade publication writers who have no intention of acknowledging the truth or anything else but their own interests.  

 

 

Talk to the technical people instead of the execs, who are repeating what they think they understand.  The streaming companies are not directly dealing with scale. They are dealing with not wanting to pay for what they use. A completely different kind of problem.

 

 

 

Anyone who considers protocol unimportant has never dealt with a cat DAC.

Robert A. Heinlein

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...