I would also have some serious concerns about "from the same mastered source".
For example, you may assume that the Red Book CD layer, and the DSD layer, on a certain hybrid SACD come from the same master...
But you have no real assurance that they weren't either processed differently - or deliberately tweaked to sound slightly different.
Likewise, I would not necessarily believe that copies of downloads, purchased or downloaded in different formats, haven't been "individually adjusted" either.
At a minimum, for the most accurate comparison, I would want to "create" the two copies myself, using a known high-quality conversion program.
Note that every conversion tends to result in a tiny measurable difference... although often not an audible one.
(This is even more true for PCM and DSD... since the formats themselves are so different a "bit perfect" conversion in either direction is not actually possible.)
So the fairest comparison would be to start with something at really high quality...
Then, using a high quality conversion program, convert it to both 24/96k PCM and DSD, and compare those.
That way you know that both started from the same "master", and both have been through a single conversion, using the same conversion program.
This is probably as close as an end user can ever come to a proper "comparison of equals".
(Otherwise you're just basing your conclusions on your best guess about what you believe the original probably sounded like.)
And, when it comes to things like MQA, any such bets are off.
MQA "encoding" involves both a form of not-quite-lossless encoding AND a form of post processing (intended to remove flaws added during the mastering process).
While the descriptions provided tend to conflate the two they are actually two entirely different things.
First, MQA claims to be able to deliver an accurate reproduction of an original, which could in principle be easily verified.
Second, MQA claims to be able to improve the sound of the master by processing it and "correcting" certain deficiencies...
(However, obviously, in order to do so, it must change the sound...)
In the context of that second claim, if the MQA processed track doesn't noticeably different than the original, then the MQA processing was wasted.
But, if it does change the sound, then the result is no longer true to the original, and we're back to a subjective judgment about "which one sounds better".
Citing
www.2l.no/hires/ as an example of MQA encoding also brings up a few interesting questions....
MQA "encoding" involves both a form of not-quite-lossless encoding
AND a form of post processing intended to remove flaws added during the mastering process.
These two things could be considered separately... although the promoters of MQA do their best to conceal that option.
(Presumably, if the recording was mastered using MQA, then those flaws would be absent, and the post processing would be unnecessary.)
If you think carefully, this process comparison is not symmetrical, and results in several different questions....
Will a track both mastered and played in MQA sound subjectively better than one mastered and played in PCM?
Will a track both mastered and played in MQA be objectively closer to the original performance than one mastered and played in PCM?
Will a track mastered in PCM, then subjected to MQA post processing, sound different (and, if so, will it sound better)?
(Most people I know agree that they often do sound difference... but not everyone agrees that the difference is always an improvement.)
If the MQA post processing produces a benefit, then can we convert the result back to PCM after we apply the post-processing, and deliver it as ordinary PCM, without losing that benefit?
Assuming that the difference is clearly audible then there is no reason to believe that it will disappear if we convert that file back into PCM once it's been "fixed".
However, the folks who developed MQA have done their best to make it impossible for us to separate "MQA processing" and "MQA encoding and decoding", so we can hear for ourselves.
They COULD offer us samples of the same file, in plain PCM, in PCM with MQA post-processing applied, and in MQA with MQA post-processing applied, so we could compare the three options...
Or they COULD offer an "MQA post-processing engine" that is separate from "the MQA encoding and decoding CODEC"...
Instead, they insist on combining the two, which sort of suggests that they're hoping that we'll eventually "just look for the logo and not sweat the details".
After that long winded explanation....
Let me conclude by saying that I personally do NOT find the sound of MQA post-processing compelling....
Of the samples I've heard, some sound a little better, some sound a little worse, some just sound a bit different, and some sound about the same....
(Which is about the same score I would grant to any of the other recent non-MQA "reissues" and "remasters" I've encountered.)
When I do this comparison I take a native DSD128 file and a PCM192k from the *same mastered source*.
Yes, judging what "sounds better" in this range is *very* subjective. That's why you can use THD/SNR/DNR as purely objective terms to "start the discussion".
You'd have to define why you say these other sources have way better SQ than a Sonica. Is this an emperical observation or are you comparing THD/SNR/DNR?
To my ear, the 9308PRo in the Sonica sounds cleaner, sounds better and is overall more pleasing.
I must question the accuracy of the 140dB assertion on Wikipedia. 140dB SPL is the onset of pain. All of the tesearch I've read states 118-119dB of dynamic range of human hearing. this is why 120dB is considered transparent.
This is a good source for content for comparisons (including MQA).
www.2l.no/hires/