MQA is Vaporware

[mansr]

2 hours ago, esldude said:

I'll shut up now.  Makes no difference anyway.  The sound we can get from downloads or CDs is what it is regardless of how it got that way. 

It's still interesting due to the weird differences between the regular and MQA versions of the album. The simplest explanation is that they are different transfers from analogue tape, and we're seeing the effects of slightly different ADC clocks and fluctuations in tape speed. If no analogue master tape was involved, something much stranger has been done.

[Jud]

2 hours ago, mansr said:

It's still interesting due to the weird differences between the regular and MQA versions of the album. The simplest explanation is that they are different transfers from analogue tape, and we're seeing the effects of slightly different ADC clocks and fluctuations in tape speed. If no analogue master tape was involved, something much stranger has been done.

 

Either someone did a lot of work just for an MQA master, or the MQA came from one of the several existing masters.  I wonder whether, if we were to do a little more research into the various versions listed on Discogs, we would find the master MQA used among them.

 

Track lengths for SACD, DVD-A, and SHM-CD versions might show something.

[mansr]

OK, let's take this recording apart. Again looking at track 2 from the Nightfly album, we begin by aligning the starts of the two version and look at the spectrogram of the left channel for the first 20 seconds, Tidal "hifi" on top and decoded MQA below:

The MQA version starts a full 8 seconds earlier, and there is a slight difference at 8-9 seconds. On closer inspection, The first second or so of the files match, then the CD is missing a chunk. Apparently the MQA process can restore content that has been edited out. Very impressive.

 

At the very end of the track, we find this:

The fadeout is clearly different. Less obvious is that here the MQA version is actually lagging by about 20 ms. In other words, it is slightly slower. It should be noted that this difference is far too small to be audible.

 

Now we trim the start and end, leaving only the part containing matching audio and align the beginning. Then for each second of audio, calculate the cross correlation between the two files and plot the lag corresponding to the maximum correlation:

It's a squiggly line. The lag with strongest correlation is increasingly negative which means the MQA file is falling behind. The overall linear nature is consistent with this being a different transfer from analogue using an ADC with a slightly faster clock. No normal digital workflow would introduce such a drift. To get a better look at the squiggles, we do a linear least squares fit and plot the residual:

This shows the relative time difference between the files for each second with compensation for the linear drift. The sudden jump at 182 seconds is a bit odd.

 

Regardless of the production details that have been debated here, it is clear that these two files come from different masters. Attributing differences in sound solely to MQA would be a big mistake.

[Jud]

7 minutes ago, mansr said:

The sudden jump at 182 seconds is a bit odd.

 

Restoring content that was edited out?

[PeterSt]

There's some DVDA version around (48/24). The second track on that version is 3:49.01 (the MQA is 3:49.41).

(the 16/44.1 I have is 3:42.46).

Maybe you knew about the DVDA version already.

 

Anyway, here you have the spectrograph of that one :

 

It is essentially the same as the MQA one.

 

I also have a 2-volume version which is 24/48 and on that one track 02 is 3:49.38.

 

Lastly I have a 24/176.4 SACD rip. Length is 3:49.53 with so-called (?) DSD mastering :

 

 

As usual, most hires is faked.

But this in itself should imply that there's a 16/44.1 around of this length. I looked, but I don't have it. Tidal neither.

[mansr]

49 minutes ago, PeterSt said:

There's some DVDA version around (48/24). The second track on that version is 3:49.01 (the MQA is 3:49.41).

(the 16/44.1 I have is 3:42.46).

Maybe you knew about the DVDA version already.

 

Anyway, here you have the spectrograph of that one :

 

If that's a 48 kHz file, why does your spectrogram go to 60 kHz?

[PeterSt]

Just now, mansr said:

If that's a 48 kHz file, why does your spectrogram go to 60 kHz?

 

It's a general routine in the software that's independent of sampling rate. So I fixed it to something "max" which can be of interest. It was actually made to judge the realness of MQA hires (for myself).

You can see that the 176.4 file also is cut 60KHz (should be 88.2 obviously).

[mansr]

8 minutes ago, PeterSt said:

It's a general routine in the software that's independent of sampling rate. So I fixed it to something "max" which can be of interest. It was actually made to judge the realness of MQA hires (for myself).

You can see that the 176.4 file also is cut 60KHz (should be 88.2 obviously).

So you resample everything to 120 kHz? Odd.

[PeterSt]

2 minutes ago, mansr said:

So you resample everything to 120 kHz? Odd.

 

If SoX is doing that during the process, then yes. But I don't do it.

I must notice that there has been a time that I explicitly wanted this behavior in the Spectrograph, but SoX couldn't do it that I could find (this is quite some years ago). This time though, it suddenly worked and I thought "okayyy, they have changed something".

But I was also working with the display resolution (via SoX) so it was more easy to compare different sampling rates. I mean, a 22.05 (44.1 sampling) of the same height as a 44.1 (88.2 sampling) doesn't really work if you want to observe how a 44.1 sampled file expands when it is upsampled (pun) to 88.2. I mean, the part up to 22.05 should stay the same and only the expansion above it counts.

So it is a little aid this way, for me.

[mansr]

56 minutes ago, PeterSt said:

If SoX is doing that during the process, then yes. But I don't do it.

I must notice that there has been a time that I explicitly wanted this behavior in the Spectrograph, but SoX couldn't do it that I could find (this is quite some years ago). This time though, it suddenly worked and I thought "okayyy, they have changed something".

But I was also working with the display resolution (via SoX) so it was more easy to compare different sampling rates. I mean, a 22.05 (44.1 sampling) of the same height as a 44.1 (88.2 sampling) doesn't really work if you want to observe how a 44.1 sampled file expands when it is upsampled (pun) to 88.2. I mean, the part up to 22.05 should stay the same and only the expansion above it counts.

So it is a little aid this way, for me.

Well, something has obviously resampled the data in the spectrogram above if the input was 48 kHz. Otherwise you wouldn't have those bursts (probably caused by clipping) reaching all the way to the top.

[mcgillroy]

On 7/22/2017 at 6:24 PM, mansr said:

The regular and MQA versions of The Nightfly on Tidal are clearly from different masters. Comparing track 2, there are a number of differences:

• The sample rates differ, 44.1 kHz vs 48 kHz for the MQA version.
• The MQA version is 7 seconds longer.
• The MQA version is 2 dB louder, even clipping a few times.
• The polarity of the MQA version is inverted.
• The actual speeds don't quite match. In other words, they are not simply different digital downsamples from a common source.
• The speed difference fluctuates throughout the track.

I'd be inclined to say these two versions came from different tape transfers, except it's an all-digital production.

This is hilarious!

[esldude]

10 hours ago, mansr said:

OK, let's take this recording apart. Again looking at track 2 from the Nightfly album, we begin by aligning the starts of the two version and look at the spectrogram of the left channel for the first 20 seconds, Tidal "hifi" on top and decoded MQA below:

The MQA version starts a full 8 seconds earlier, and there is a slight difference at 8-9 seconds. On closer inspection, The first second or so of the files match, then the CD is missing a chunk. Apparently the MQA process can restore content that has been edited out. Very impressive.

 

At the very end of the track, we find this:

The fadeout is clearly different. Less obvious is that here the MQA version is actually lagging by about 20 ms. In other words, it is slightly slower. It should be noted that this difference is far too small to be audible.

 

Now we trim the start and end, leaving only the part containing matching audio and align the beginning. Then for each second of audio, calculate the cross correlation between the two files and plot the lag corresponding to the maximum correlation:

It's a squiggly line. The lag with strongest correlation is increasingly negative which means the MQA file is falling behind. The overall linear nature is consistent with this being a different transfer from analogue using an ADC with a slightly faster clock. No normal digital workflow would introduce such a drift. To get a better look at the squiggles, we do a linear least squares fit and plot the residual:

This shows the relative time difference between the files for each second with compensation for the linear drift. The sudden jump at 182 seconds is a bit odd.

 

Regardless of the production details that have been debated here, it is clear that these two files come from different masters. Attributing differences in sound solely to MQA would be a big mistake.

Did not some early Sony gear run at 44,101 hz? Or am I mis-remembering that?  That would be near a 23 ppm speed difference which also would look like a minor clock speed difference.  That might explain your first graph though not the second.  If they were using an early Sony CD master for one of the versions, and that version was done on 44,101 hz sampling, while an MQA version ran at the proper though slightly slower speed of 44,100 hz. 

 

Its not clear without labels what your graph is showing.  Percent or milliseconds or sample slippage or what nor over what recording time you used.  So I don't know if that result fits the 23 ppm speed differences or not.

[mansr]

12 minutes ago, esldude said:

Did not some early Sony gear run at 44,101 hz? Or am I mis-remembering that?  That would be near a 23 ppm speed difference which also would look like a minor clock speed difference.  That might explain your first graph though not the second.  If they were using an early Sony CD master for one of the versions, and that version was done on 44,101 hz sampling, while an MQA version ran at the proper though slightly slower speed of 44,100 hz. 

The average difference here is 92 ppm with the MQA version having more samples per seconds. How that came to be might be interesting, but whatever the explanation, it doesn't alter the fact that different masters are involved. The selling point of MQA (presented to consumers) is that of "de-blurring." Unless this process can magically restore 8 seconds of deleted audio, we're looking at something else entirely.

[mansr]

45 minutes ago, esldude said:

Its not clear without labels what your graph is showing.  Percent or milliseconds or sample slippage or what nor over what recording time you used.  So I don't know if that result fits the 23 ppm speed differences or not.

Sorry, all axes are labelled in seconds.

[esldude]

1 hour ago, mansr said:

The average difference here is 92 ppm with the MQA version having more samples per seconds. How that came to be might be interesting, but whatever the explanation, it doesn't alter the fact that different masters are involved. The selling point of MQA (presented to consumers) is that of "de-blurring." Unless this process can magically restore 8 seconds of deleted audio, we're looking at something else entirely.

I suppose that highlights the question that doesn't get asked.

 

MQA=Master Quality Authenticated

 

Nobody asks "Which master?"

[Ralf11]

it's recursive - MQA is the Master

 

once DRM is propagated all over Skynet

[esldude]

13 minutes ago, Ralf11 said:

it's recursive - MQA is the Master

 

once DRM is propagated all over Skynet

Would have been more obvious is they named it MQAQM.

 

Master Quality Authenticated Quality Master

[Digital Assassin]

29 minutes ago, esldude said:

I suppose that highlights the question that doesn't get asked.

 

MQA=Master Quality Authenticated

 

Nobody asks "Which master?"

Let's be clear.

 

NOTHING is being authenticated, WHAT SO EVER.

 

This is is one of the most laughable outright lies told by MQA from the beginning...that somehow

the artist and producer would "sign off" on the master files processed by Meridian.

 

The only thing that is authenticated is that one is a sucker and bought an MQA DAC.

[Sal1950]

4 hours ago, Digital Assassin said:

The only thing that is authenticated is that one is a sucker and bought an MQA DAC.

That and the fact that he gets to see the little green "Authenticated" light illuminate.

Isn't he the lucky soul.  LOL

[PeterSt]

8 hours ago, mansr said:

Unless this process can magically restore 8 seconds of deleted audio, we're looking at something else entirely.

 

You are serious, right ?

[soxr]

7 hours ago, esldude said:

Would have been more obvious is they named it MQAQM.

 

Master Quality Authenticated Quality Master

 

 

Master Quality Approximated ?

 

[Sal1950]

It's all about the "Deblurschitt"  

[PeterSt]

Hmm ...

 

MQA does a better job than they know themselves. Unless of course the 24 bits I listen to when decoded are actually 16 bits only.

I could also say : WTF ?

 

Does anyone know the (un)truth about this ?

I don't think I heard any talk about this, but it seems to say that only with renderer you'll receive 24 bits. Okay, FWIW of course. But the opposite "you sure will not receive more than 16 bits with software decoding only" would be a bummer.

 

And there we go again ... how to test this ...

 

 

[soxr]

6 hours ago, Sal1950 said:

It's all about the "Deblurschitt"  

At least real studio engineers authenticate MQA not to be the original studio sound, but:
 

Quote

MQA is, in all of my tests so far, brighter and thinner with more distortion, and a sense of excitement (remember Aphex?) and even volume from the artifacts. Sure many lay people will be fooled. I don't like to see pros fooled, but oh well, that's why I'm busy I suppose and they're posturing next to a new product so I'm free to tell the truth here.

https://www.gearslutz.com/board/12548751-post460.html

So MQA deblur (killing post-ringing of minimum phase resampling in the renderer phase with weird filters) is the shit sold to audiophiles as the new PCM, but it's not the studio sound.

And let's remember: all DSP applied by MQA is still in the PCM domain.
 

[Display]

On 7/25/2017 at 4:58 PM, soxr said:

At least real studio engineers authenticate MQA not to be the original studio sound, but:
 

https://www.gearslutz.com/board/12548751-post460.html

So MQA deblur (killing post-ringing of minimum phase resampling in the renderer phase with weird filters) is the shit sold to audiophiles as the new PCM, but it's not the studio sound.

And let's remember: all DSP applied by MQA is still in the PCM domain.
 

 

Technical justification is futile  - "They Are the MQA. You Will be Assimilated. Resistance is Futile".