MQA The Truth lies Somewhere in the Middle

[John_Atkinson]

1 hour ago, adamdea said:

it isn't a series of the things you used in that article either.

 

I think you are missing the point of that article. I started with an analog signal with a  60kHz bandwidth. Conventional Shannon-type A/D conversion of that signal to 96kHz and conventional sample-rate conversion to 44.1kHz both introduced of Nyquist frequency ringing when musical transients arose from digital black, the audio equivalent of "edges" in image processing. This ringing at Nyquist stimulates identical ringing in a DAC's conventional reconstruction filter.

 

John Atkinson

Editor, Stereophile

[John_Atkinson]

1 hour ago, Albrecht said:

echos much of what Charles Hanson has had to say...

 

Hansen. How difficult is it for you to spell his name correctly?

 

John Atkinson

Editor, Stereophile

[John_Atkinson]

8 minutes ago, crenca said:

You owe Albrecht an apology.

 

I hardly think so. Charley Hansen is dead. The least people can do is pay him the respect of spelling his name correctly.

 

John Atkinson

Editor, Stereophile

 

[John_Atkinson]

4 minutes ago, Albrecht said:

Corrected my spelling error of Charles' name. I have been/am a fan of Charles' work and would never intentionally disparage his work, ideas, or person.

 

Thank you.

 

John Atkinson

Editor, Stereophile

 

[John_Atkinson]

On 10/15/2018 at 5:45 PM, John_Atkinson said:

When I have a  spare minute, I will take a look at the difference between the channels of both files, which should reveal any truth in Mr. Carver's assertion.

 

Bob Carver claimed that the MQA files sent to writers had been processed with something like his Sonic Holography circuit. This circuit is basically a Blumlein "shuffler," which splits the stereo audio signal into its Sum and Difference components, applies gain and/or EQ to the Difference component, then recombines the Sum and Difference components into Left and Right signals.

So, if Carver's claim is correct, then examining the Difference between the two channels of an MQA file and the PCM original will reveal if any "shuffling" had been done.

I took 3 recordings for which I had both the originals and the MQA versions that I had used for my initial auditioning of MQA back in 2016 where I had found the MQA version to sound superior: a track from Keith Jarrett's Cologne Concert, a Debussy prelude, and one of my choral recordings with the Portland State Chamber Choir.

Extracting the difference information for the PCM originals was straightforward. For the MQA versions, I decoded the files with Roon feeding my PS Audio DAC and and digitized the PS Audio's analog out with an Ayre A/D converter operating at 24 bits and 192kHz sample rate. I adjusted the peak levels of the resultant files to match those of the original PCM files and extracted the channel difference information.

In all 3 cases, the channel-difference file of the MQA version had the same peak level as that of the original PCM file, as well as virtually identical spectra. So Bob Carver was wrong.

John Atkinson
Editor, Stereophile

[John_Atkinson]

14 hours ago, Miska said:

15 hours ago, John_Atkinson said:

Extracting the difference information for the PCM originals was straightforward.

 

That doesn't sound straightforward at all.

 

By straightforward, I meant that it was conceptually simple: load the file into a DAW; subtract the right channel from the left channel, save the resultant mono information; and analyze those data for spectrum and peak and average levels.

 

14 hours ago, Miska said:

It really shouldn't be that difficult. And it is not, better make digital capture of the decode for comparisons.

 

That is what I did for the MQA-encoded files, matching the peak levels of the digital capture of the MQA playback to those of the original files. The MQA files were all unfolded to 2Fs so by running the A/D converter at 4Fs, its own behavior would be out-of-band. And when those data were analyzed, the differences between the channels were the same as those of the original PCM files, meaning that Bob Carver's unsupported assertion that MQA encoding includes crosstalk cancellation - he didn't perform difference tests - was wrong.

 

John Atkinson

Editor, Stereophile 

[John_Atkinson]

1 minute ago, mansr said:

You said you did an analogue capture.

 

By "digital" I mean that the decoded MQA analog signals were redigitized for analysis.

 

MQA files played with Roon server -> network connection -> PS Audio DAC with volume control set to its maximum, thus bypassed -> analog output -> Ayre QA-9 ADC at 192kHz -> USB connection -> Host PC.

 

John Atkinson

Editor, Stereophile

[John_Atkinson]

1 hour ago, mansr said:

It would help the discussion if you used words in accordance with their established meanings.

 

In my original posting, I wrote "For the MQA versions, I decoded the files with Roon feeding my PS Audio DAC and and digitized the PS Audio's analog out with an Ayre A/D converter operating at 24 bits and 192kHz sample rate."
 

I don't see what's unclear about that.

 

John Atkinson

Editor, Stereophile

 

[John_Atkinson]

47 minutes ago, Miska said:

there's unnecessary analog path on the way...

 

Even so, the fact that the difference files from the decoded MQA files when converted from analog to digital so closely matched the difference files from the original 2Fs PCM data indicates that Bob Carver's statement was incorrect.

 

For example, here are the amplitude statistics for the difference data with the Debussy Prelude recording:

Debussy original 96kHz L-R difference file:

Peak Amplitude -5.23dB

Total RMS -29.06 dB

ITU Loudness -31.12 LUFS

Debussy decoded MQA L-R difference file, redigitized at 192kHz:

Peak Amplitude - 5.22dB

Total RMS -29.03 dB

ITU Loudness -31.13 LUFS

 

No evidence for crosstalk cancellation.

 

John Atkinson

Editor, Stereophile

[John_Atkinson]

1 minute ago, Rt66indierock said:

 There were MQA files floating around in 2014 and 2015.  Is Bob [Carver] talking about these files? 

 

That's what I was told.

 

John Atkinson

Editor, Stereophile

[John_Atkinson]

6 minutes ago, Rt66indierock said:

And could the MQA files given to you be post this this period and processed differently? 

 

No, as Bob Carver was clearly talking about the first set of MQA files that had been sent to journalists and industry people. A colleague had sent Mr. Carver some of those files, so as I had still had the same set of files in my DropBox folder, I performed the analyses mentioned earlier on 3 of them that I had used for my listening comparisons that were reported in Stereophile.

 

John Atkinson

Editor, Stereophile

[John_Atkinson]

On 11/7/2018 at 5:52 PM, Magnaryder said:

I guess a little background is required to reference Mr. Atkinson's remarks about Bob Carver. 

 

For those unaware of the body of work
 

Carver caused a stir in the industry in the mid-1980s when he challenged two high-end audio magazines to give him any audio amplifier at any price, and he’d duplicate its sound in one of his lower cost (and usually much more powerful) designs. Two magazines accepted the challenge. We'll concern ourselves with Stereophile

 

The objective was to make the two amplifiers sound absolutely identical, or at least similar enough in sound that none of us could tell one from the other with better than 50% (pure chance) consistency.
Read more at https://www.stereophile.com/content/carver-challenge#BGvbTXCZstGTCj1u.99

 

The Stereophile employees failed to pass a single blind test with their own equipment in their own listening room.

 

from the Challenge: "After the second day of listening to his final design, we threw in the towel and conceded Bob the bout." 
Read more at https://www.stereophile.com/content/carver-challenge-page-4#gjfGTsvQog7G6iAU.99

 

 

Thank you for linking to the article. However, I am not sure why you are linking it with my comments on Bob Carver's claim that the first set of MQA-encoded files were processed with something similar to his "Sonic Holography" circuit. This "Carver Challenge" took place in 1985, before I joined Stereophile. The Stereophile "employees" who took part in these listening tests, using, I was later told, the treble and midrange panels of a pair of Infinity RS-1B speakers, were then owner Larry Archibald and editor/founder J. Gordon Holt.

 

John Atkinson

Editor, Stereophile

[John_Atkinson]

27 minutes ago, Rt66indierock said:

They pay attention to what the members of what John Atkinson calls “the we oppose MQA fraternity” ...

 

I must be getting forgetful in my old age, but I don't remember saying that? Link?

 

And on a more important subject, Hilary Hahn's recording of the Elgar Violin Concerto is definitive,

https://www.stereophile.com/content/2008-records-die-page-2

 

John Atkinson

Editor, Stereophile

[John_Atkinson]

42 minutes ago, kumakuma said:

 

 

Thanks. I must indeed be getting forgetful. Must have been listening to too much music encoded with MQA.  ?

 

John Atkinson

Editor, Stereophile

 

[John_Atkinson]

1 minute ago, kumakuma said:

Happens to us all...

 

What does?

 

John Atkinson

Editor, Stereophile