MQA vs HiRez: an apples-to-apples comparison

[manisandher]

5 minutes ago, mansr said:

I'm talking about what happens after that. The missing renderer instructions mean that something has altered the data after it left the MQA decoder.

 

I doubt Roon would send any renderer instructions if 'no MQA support' is selected... as indeed it was.

 

Mani.

[mansr]

7 minutes ago, manisandher said:

I doubt Roon would send any renderer instructions if 'no MQA support' is selected... as indeed it was.

Meaning it must have stripped them from the MQA decoder output. Who is to say it didn't do this in a way that had audible side effects? As long as there is an unknown processing stage involved, any comparison is meaningless.

[manisandher]

1 minute ago, mansr said:

Meaning it must have stripped them from the MQA decoder output. Who is to say it didn't do this in a way that had audible side effects? As long as there is an unknown processing stage involved, any comparison is meaningless.

 

So the fact that Roon optimises MQA playback for non-MQA DACs makes this comparison meaningless? How convenient for you not to have to find the 'obvious ways in which MQA molests the pure hirez'.

 

1 hour ago, mansr said:

If you want an accurate comparison, the first step is to make sure you haven't accidentally altered the files in some way that wouldn't happen on normal playback.

 

I've played the MQA file back exactly as anyone with a non-MQA DAC would using Roon. It's not my fault Roon have got their act together.

 

Mani.

[manisandher]

8 minutes ago, mansr said:

Who is to say it didn't do this in a way that had audible side effects?

 

Again...

 

This would make it even easier to identify the MQA from the pure, unadulterated hirez. So, which is it?

 

Mani.

[mansr]

How can we be sure Roon hasn't applied its own sauce to the non-MQA file as well?

[manisandher]

18 minutes ago, mansr said:

How can we be sure Roon hasn't applied its own sauce to the non-MQA file as well?

 

Why would it? To bring the quality of all files down to the extent that unadulterated hirez is indistinguishable from 'totally molested' MQA?

 

Mani.

 

[manisandher]

OK, I can't make it any easier than this. Here's the original 24/88.2 (for a limited time only, because I don't want to get into trouble):

 

https://drive.google.com/open?id=1jAdGtBzHDMnDlX0O3ArdhczdRGDkCMpG

 

It should be really trivial to identify the 'totally molested' MQA now.

 

Knock yourself out...

 

Mani.

[PeterSt]

1 hour ago, manisandher said:

Roon is in Shared Mode (correct Peter). Audacity is capturing via WASAPI.

 

Yes but ... 

[wklie]

53 minutes ago, mansr said:

How can we be sure Roon hasn't applied its own sauce to the non-MQA file as well?

 

Generally, with DSP engine disabled, volume leveling disabled, and in single-zone playback, Roon will not do anything to change the music data other than to accommodate DAC limitations (e.g. ancient 44.kHz-only DAC).  Any processing Roon does to music data will show up in the signal path.  If Roon says it's doing lossless sending of data, it is doing lossless sending of data.  I believe somebody tested that using the special PS Audio-specific lossless test in the past.

[manisandher]

3 minutes ago, wklie said:

 

Generally, with DSP engine disabled, volume leveling disabled, and in single-zone playback, Roon will not do anything to change the music data other than to accommodate DAC limitations (e.g. ancient 44.kHz-only DAC).  Any processing Roon does to music data will show up in the signal path.  If Roon says it's doing lossless sending of data, it is doing lossless sending of data.  I believe somebody tested that using the special PS Audio-specific lossless test in the past.

 

It's says it's doing lossless sending of data:

 

 

 

Now whether Shared Mode is having an affect...

 

Mani.

[pkane2001]

58 minutes ago, manisandher said:

OK, I can't make it any easier than this. Here's the original 24/88.2 (for a limited time only, because I don't want to get into trouble):

 

https://drive.google.com/open?id=1jAdGtBzHDMnDlX0O3ArdhczdRGDkCMpG

 

It should be really trivial to identify the 'totally molested' MQA now.

 

Knock yourself out...

 

Mani.

 

My analysis:   D = original PCM, C = MQA

 

Here's the delta spectrum of the original minus D:

Volume diff= 0.000dB  Phase offset=701.09ms (61836.000 samples)
Difference (rms) = -132.30dB
Correlated Null Depth=202.93dB
Phase drift before correction: 0.0000 ppm

 

And here's original minus C :

Volume diff= -0.003dB Phase offset=644.02ms (56802.890 samples)
Difference (rms) = -54.84dB
Correlated Null Depth=62.84dB
Phase drift before correction: -0.0007 ppm

 

[church_mouse]

I have no idea about the technicalities.  I have been listening to C v D much of today using a shuffled playlist and I have preferred D every time.  

 

Perhaps, in due course, someone will tell me which is which.

[mansr]

6 minutes ago, pkane2001 said:

My analysis:   D = original PCM, C = MQA

 

Here's the delta spectrum of the original minus D:

If the capture were perfect, without any alteration, it should have been an exact match. Something is messing with the data.

[manisandher]

1 minute ago, church_mouse said:

I have no idea about the technicalities.  I have been listening to C v D much of today using a shuffled playlist and I have preferred D every time.  

 

Perhaps, in due course, someone will tell me which is which.

 

Excellent! I certainly will let you know... in due course.

 

If it's possible for you to articulate, what do you prefer about D over C? (I realise these things aren't easy.)

 

Mani.

[manisandher]

12 minutes ago, pkane2001 said:

 

My analysis:   D = original PCM, C = MQA

 

Here's the delta spectrum of the original minus D:

Volume diff= 0.000dB  Phase offset=701.09ms (61836.000 samples)
Difference (rms) = -132.30dB
Correlated Null Depth=202.93dB
Phase drift before correction: 0.0000 ppm

 

And here's original minus D :

Volume diff= -0.003dB Phase offset=644.02ms (56802.890 samples)
Difference (rms) = -54.84dB
Correlated Null Depth=62.84dB
Phase drift before correction: -0.0007 ppm

 

 

Thanks Paul (as always). But you have two 'original minus D' there. Which plot is which?

 

Edit: I can see that the second plot is 'original minus C'.

 

Mani.

[PeterSt]

8 minutes ago, pkane2001 said:

And here's original minus D :

 

13 minutes ago, pkane2001 said:

My analysis:   D = original PCM, C = MQA

 

 

[manisandher]

3 minutes ago, PeterSt said:

 

 

 

 

You're a man of very few words today Peter!

 

Mani.

[pkane2001]

14 minutes ago, manisandher said:

 

Thanks Paul (as always). But you have two 'original minus D' there. Which plot is which?

 

Edit: I can see that the second plot is 'original minus C'.

 

Mani.

 

Yep, sorry, last minute edit messed that up. First one is the original - D.

[manisandher]

1 hour ago, manisandher said:

OK, I can't make it any easier than this. Here's the original 24/88.2 (for a limited time only, because I don't want to get into trouble):

 

https://drive.google.com/open?id=1jAdGtBzHDMnDlX0O3ArdhczdRGDkCMpG

 

It should be really trivial to identify the 'totally molested' MQA now.

 

Knock yourself out...

 

Mani.

 

I've taken the original hirez file down as @pkane2001's software has performed its magic.

 

Mani.

[manisandher]

6 hours ago, Miska said:

By quick glance A [C] looks like MQA and B [D] like original. 

 

32 minutes ago, pkane2001 said:

My analysis:   D = original PCM, C = MQA

 

Well done chaps. Your analysis skills are exemplary.

 

29 minutes ago, church_mouse said:

I have been listening to C v D much of today using a shuffled playlist and I have preferred D every time.

 

And a special well done to you Sir (or Madam)!

 

Yes, C is the decoded MQA and D the original hirez.

 

I hear a clear difference between the two myself, but have to say that there remains something very 'nice' about the way C sounds. Due to 'deblur' or added distortion? I don't know.

 

Although the cat's now out of the bag, there's still no reason not to listen to the two files and hear the differences for yourself... if you feel so inclined.

 

Mani.

[crenca]

48 minutes ago, wklie said:

 

Generally, with DSP engine disabled, volume leveling disabled, and in single-zone playback, Roon will not do anything to change the music data other than to accommodate DAC limitations (e.g. ancient 44.kHz-only DAC).  Any processing Roon does to music data will show up in the signal path.  If Roon says it's doing lossless sending of data, it is doing lossless sending of data.  I believe somebody tested that using the special PS Audio-specific lossless test in the past.

 

Except with MQA it is not pure because it is applying a DSP process to even get at the data.... Another way of saying that is there is no such thing as "lossless" when processing MQA.  As I think yourself and mansr are pointing to the LSB is being stripped...

[church_mouse]

Mani, I feared you would ask that.

 

On D the singer’s voice seemed clearer, with start and end of words more precise.  The percussion also seemed better defined.  One track, which turned out to be D each time, I was hearing as less “fuzzy” than the other (a relative description, rather than an absolute statement that C is fuzzy).  I also felt C came out at me more when listening through speakers, but at the expense of background instrumental detail and with a comparatively false bass enhancement.

 

If you now tell me the two tracks are identical I fear there may be reports of “man seen jumping into river in Stratford upon Avon shouting “I know nothing!””.

[mansr]

3 minutes ago, church_mouse said:

On D the singer’s voice seemed clearer, with start and end of words more precise.  The percussion also seemed better defined.  One track, which turned out to be D each time, I was hearing as less “fuzzy” than the other (a relative description, rather than an absolute statement that C is fuzzy).  I also felt C came out at me more when listening through speakers, but at the expense of background instrumental detail and with a comparatively false bass enhancement.

Perhaps one of the tracks might be described as somewhat blurry?

[manisandher]

4 minutes ago, church_mouse said:

Mani, I feared you would ask that.

 

On D the singer’s voice seemed clearer, with start and end of words more precise.  The percussion also seemed better defined.  One track, which turned out to be D each time, I was hearing as less “fuzzy” than the other (a relative description, rather than an absolute statement that C is fuzzy).  I also felt C came out at me more when listening through speakers, but at the expense of background instrumental detail and with a comparatively false bass enhancement.

 

If you now tell me the two tracks are identical I fear there may be reports of “man seen jumping into river in Stratford upon Avon shouting “I know nothing!””.

 

Followed by "man rescued by Leamingtonian". It seems we're neighbours... relatively speaking!

 

Fear not, your ears are obviously good.

 

Mani.

[manisandher]

1 hour ago, pkane2001 said:

And here's original minus C :

Volume diff= -0.003dB Phase offset=644.02ms (56802.890 samples)
Difference (rms) = -54.84dB
Correlated Null Depth=62.84dB
Phase drift before correction: -0.0007 ppm

 

 

Does anyone have any idea of what's causing this?

 

Is this the effect of 'deblurring' in action? ('Inaction'? ?)

 

There's something about the sound of C that I really like, and I'd like to understand what it could be...

 

Mani.