MQA is Vaporware

[vortecjr]

Here is the same file streamed from my Mytek DAC to my analyzer with MQA disabled. 96kHz content according to J-River, 96kHz output according to Mytek DAC. The Mytek DAC does not have a purple light next to the MQA logo. Here is the output of my analyzer:

[vortecjr]

10 minutes ago, mansr said:

Can you increase the bandwidth of the spectrum display to 100 kHz?

Here you go with MQA enabled at the DAC:

[asdf1000]

5 minutes ago, vortecjr said:

The Benchmark DAC does not have a purple light next to the MQA logo.

 

Which model Benchmark DAC? They support MQA rendering?

 

[vortecjr]

13 minutes ago, mansr said:

Can you increase the bandwidth of the spectrum display to 100 kHz?

Here you go with MQA disabled at the DAC:

[vortecjr]

18 minutes ago, Em2016 said:

 

Which model Benchmark DAC? They support MQA rendering?

 

Sorry. I meant Mytek Brooklyn DAC with firmware 2.36 rev 0654. 

[mansr]

6 minutes ago, vortecjr said:

Here you go with MQA disabled at the DAC:

A linear frequency scale would be more useful here.

[mansr]

4 minutes ago, vortecjr said:

It's a Mytek Booklyn DAC with firmware 2.36 rev 0654. 

Mytek != Benchmark.

[vortecjr]

24 minutes ago, mansr said:

Mytek != Benchmark.

My bad I was swapping emails with a customer at the same time. I'll update the posts.

[asdf1000]

12 minutes ago, vortecjr said:

Sorry. I meant Mytek Brooklyn DAC with firmware 2.36 rev 0654. 

 

Cool.

 

Was confused as Benchmark don't have any MQA DAC.

[mansr]

1 minute ago, Em2016 said:

Cool.

 

Was confused as Benchmark don't have any MQA DAC.

Also explains the silly amount of harmonic distortion. Benchmark should be a lot better than that.

[vortecjr]

 

39 minutes ago, mansr said:

A linear frequency scale would be more useful here.

Here you go with MQA enabled at the DAC and linear scale on the X-axis:

[vortecjr]

I updated the graph on post #16987 to show dBrA on the Y-axis. Not sure if you need to refresh the webpage to see it. 

[John_Atkinson]

2 hours ago, mansr said:

Something here doesn't add up. The spectrum clearly shows a -30 dB component at 86 kHz, yet the waveform looks smooth. If I calculate and plot "sin(t) + sin(8.6 t) * 10^(-30/20)" the result looks like this:

 

Quite squiggly.

 

You are correct. Eyeballing your scope trace, the 86kHz content appears to be just under 100mV p-p compared with the 2V p-p 10kHz level, which is close to -30dB. I think there must have been a low-pass filter somewhere in my waveform shots, though not the spectra. I will redo the tests both with your original file and with the new one at -3dBFS.

 

John Atkinson

Technical Editor, Stereophile

 

[manisandher]

The 86kHz 'noise' is just the imaging (48+38 kHz) due to the 'leaky' MQA filter. I'm sure MQA would argue that no regular music content would have any signal lying at -10dB at 10kHz.

 

I'm sure something like the 'polynomial' filters in HQPlayer would produce a similar result with a -10dB 10kHz signal. Probably why @Miska recommends against using them.

 

Mani.

[John_Atkinson]

1 hour ago, John_Atkinson said:

I think there must have been a low-pass filter somewhere in my waveform shots, though not the spectra.

 

It turned out that even though I had set the analyzer's sample rate to 200kHz, it had defaulted to 48kHz for the waveform capture. Here is a repeat of the Mytek's analog output waveform taken with the correct 200kHz sample rate. The aliased product at 86kHz (96kHz-10kHz) is now visible.

 

John Atkinson

Technical Editor, Stereophile

 

 

[mansr]

9 minutes ago, manisandher said:

The 86kHz 'noise' is just the imaging (48+38 kHz) due to the 'leaky' MQA filter.

Of course it is.

 

9 minutes ago, manisandher said:

I'm sure MQA would argue that no regular music content would have any signal lying at -10dB at 10kHz.

If the content is such that their filters don't do any damage, then it follows that they also can't do any good.

[botrytis]

So this is the "MQA ' treatment of music files? NO THANK YOU!!!!!

[Kyhl]

17 hours ago, mansr said:

Here's the spectrum (normalised so the peak hits 0 dB) of a software "render" of that signal:

The 86 kHz artefact is supposed to be there at -24 dB. Could it be that PS Audio DAC has a low-pass filter somewhere after the MQA rendering?

I might be mistaken but my understanding of the PS DirectStream DACs is that they upsample everything to DSD then run it through a low pass filter (transformer based) to create the analog output, instead of a chip DAC conversion.

[mansr]

2 minutes ago, Kyhl said:

I might be mistaken but my understanding of the PS DirectStream DACs is that they upsample everything to DSD then run it through a low pass filter (transformer based) to create the analog output, instead of a chip DAC conversion.

In that case the MQA processing happens before the DSD conversion.

[John_Atkinson]

8 minutes ago, mansr said:

In that case the MQA processing happens before the DSD conversion

 

That's correct. The MQA decoding and/or rendering is performed in the network card, which then feeds linear PCM data to the PS Audio DAC module.

 

John Atkinson

Technical Editor, Stereophile

[firedog]

Okay, are we agreed now that the MQA renderer/filters distorts the waveform?

[Ishmael Slapowitz]

somebody mentioned Brian Lucey a few pages back, and I see we have a plea to agree MQA distorts the sound..

 

Well a quick google search turned up this reply to Jim Austin's part 2 article:

 

WOW!

Music is Distortion, not Perfection. 

Submitted by Brian Lucey on August 8, 2019 - 2:27am 

I'm late to the party it seems. Working daily in the studio will do that. 

MQA is a rude, cynical business losing millions, it's a harmonic scheme for money, it's sold as lossless which is a lie (although Austin is giving cover for that early claim of a lossless patent by saying "who cares?"). MQA is sold as Mastering Engineer Authenticated, meaning approved ... which is the lie to end all lies. Also sold as "correction" which is tough to believe someone would say with a straight face. 

 

They are bulk processing back catalogs to create a market, so MQA has zero integrity, yes Bob I mean you personally. MQA processing of approved masters is altering, meaning damaging/changing/stepping on client/producer/label/manager/artist approved work to make money for these guys. No one needs it, except Stuart and the team of greedy people on board. As a mastering engineer it's offensive without words. These men at MQA lost their Meridian business to DVD. Sorry gents, that is rough, and I feel for you ... yet do not go putting your greedy, manipulative, authoritarian fingers into the Recorded History of Music with this offensive bull sharkey. 

 

Certainly there are subjective cases of "preference" for the 'Sound of MQA', because like mp3, or Mastered for iTunes or a DA or speaker ... everything has a sound. People are people, the ego likes to have a vote. MQA processing to my ear, listening to my work pre and post MQA, has some harmonic distortion and maybe even a volume boost as a result just big enough to help win an A/B. Louder is better ! 

 

I understand audiophiles can be suckered. The mastering engineers who like this artifact might as well wave a banner saying "I can't craft untouchable masters on my own, I'm inept, so I need this artifact, randomly, all over my shoddy work!" The rest of us are 100% against this travesty for profit. 

 

And please, don't talk about the vetting. Of course one division of a major label (accounting) will overrule another (content producers), that's just corporate greed. 

 

In principle, "correction" of music is a fallacy. There is no perfection in music or music playback. Rooms, temperature, humidity, we the listener, are all in flux. Mastering finds a repeatable result knowing it's in flux. 

 

Also ... what is the personal insecurity of those of us who seek perfection with music? Music is organized distortion, from room compression to the massive additive distortion happening today with everything produced except maybe classical. We add distortion to recordings, like we alter the EQ of mono tracks in a mix ... we love distortion ... for the emotional impact. And by the mastering stage, we coalesce this cocktail of artifacts with supreme precision. Everything interacts. And it's signed off on by all parties. MQA steps on all of this. 

 

Dear folks at MQA, 16 bits is not a small file. People who care will download the larger files, not stream. And faster streaming gives MQA a death date just like DVDs, did you not learn anything? If you want to make serious money and change the world, build a better mp3. We all would love that and you would make billions. 

 

Finally, and slightly off topic. 44.1 is not inferior to 88.1 or 96k or 192. Therefore getting 192 down to 48 by (insert BS term here) is not anyone's goal who understands music production. The "more samples is better" myth is built on the notion of music as perfectionism. Conversion QUALITY is four things: Analog path, clock, converter chip and filter. 44.1 can be great. 96k can be bad. It's about the hardware in total. Perfectionism and "fixing" after mastering ... could not be more naive (giving Stewart the benefit of the doubt) and thus dangerous. 

 

Music intends four things: Intimacy, Connection, Community and Elevation. There is nothing perfect needed, possible or part of the listening transaction. In fact, we like the imperfections. The humanity. Please, stop the madness, it's rude and dumb and set to die in time anyway. 

 

The ONLY FILE that matters is the native sample rate of the mastering session. It cannot be improved in any way by anything that changes it. The rest is lies for money. Like this article, many people making money here. Even from the controversy. Get a real job. Go make some art. Create something. Or at the least, don't be complicit.

 

I was contacted by Mytek to represent MQA in LA, along with Bob Ludwig who they hoped to represent on the East Coast. They processes my work in the best way they can, and I have since heard some of my work catalog post processing. Yuk. If I wanted that distorting in there, I would have added it in the first place.

https://www.stereophile.com/content/mqa-tested-part-2-fold#OuSFKFQrkGOShLt3.99

 

[MikeyFresh]

12 minutes ago, firedog said:

Okay, are we agreed now that the MQA renderer/filters distorts the waveform?

 

I'd like to think we can all agree on that one, but I'm not an engineer, so perhaps I've missed something.🤑

[John_Atkinson]

16 hours ago, John_Atkinson said:

I got different results with the Mytek Brooklyn. The front panel indicated that the 24/96 FLAC file was being rendered as 24/192k and a red rather than blue light illuminated next to the MQA indicator. This time, the spectrum with the 10kHz at 0dBFS was dirty, with an 86kHz component at -30dB. The waveform, however, was still visually clean.

 

 

Here are both the spectrum of the Mytek Brooklyn's analog output rendering mansr's original file, 10kHz at 0dBFS, which he says may have had some clipping (or inter-sample overs), and his new file, 10kHz at -3dBFS. The latter has a much cleaner spectrum, though the aliased product at 86kHz lies close to the same level.

 

John Atkinson

Technical Editor, Stereophile

[mansr]

4 minutes ago, MikeyFresh said:

I'd like to think we can all agree on that one, but I'm not an engineer, so perhaps I've missed something.

It doesn't take an engineer to tell a smooth curve from a squiggly one.