It seems to me that, contrary to usual proper test procedure, the audio tracks used for comparison should NOT be identically mastered. Subjective listening tests utilizing completely identical tracks would seem to work to obscure whatever subjective sound benefits MQA may possibly bring over RBCD or High-Rez, rather than reveal them. The MQA targeted version of test tracks should be mastered to optimize the signal time-domain, where as the RBCD and High-Rez targeted version of those same test tracks should be mastered to optimize the signal frequency-domain, which is how they are typically mastered for commercial release.
Time-domain prioritization is the primary performance objective of MQA as this reduces signal transient smearing. While the audible significance of music transient smearing in digital audio is debated, Meridian believes it is an major factor in unpleasant sounding RBCD, and to a lesser degree, High-Rez. That objective requires a format purposely engineered to prioritize the time-domain. Which includes requiring that the music be mastered to optimize the time-domain. Specifically, I'm refering to the implementation of the anti-alias filter utilized in mastering. The usual frequency-domain optimization of the channel leads to utilizing brickwall anti-alias filters (of which there is no choice with RBCD), while MQA instead utilizes a much more gently sloping anti-alias filter. The anti-alias filter slope has direct implications for the signal time-domain parameters, such as transient smearing.
By the way, while both RBCD and High-Rez are typically mastered utilizing brickwall anti-alias filters, the transient smearing via High-Rez brickwall filters is less than that of RBCD brickwall filters, due to the higher sample rate of High-Rez. The higher rate shortens the duration of the transient smearing, and might be the primary technical reason why High-Rez sounds superior (if only marginally) to RBCD.
Potentially, further complicating achieving an valid comparison is that unless those who are conducting the comparison know exactly how the test tracks were mastered, it's possible that any High-Rez test tracks were, intentionally or not, mastered optimizing the time-domain, thereby delivering the primary performance benefit of MQA, and with less loss (quantization noise) too. That also highlights the real performance benefit offered by MQA, which is that it offers a time-domain prioritized standard/requirement for mastering (and playback). It does not offer some new kind of time-domain optimizing DSP technology. This is an important distinction.
So, while High-Rez is capable of being mastered for the time-domain there isn't and industry requirement for that (to my limited outsider's knowledge) via High-Rez. RBCD must utilize brickwall anti-alias filters because the signal band extends close enough to the channel Nyquist frequency to otherwise risk signal aliasing without them. MQA does additionally include the DAC in it's time-domain prioritizated system strategy, which is an element of the system chain left out of being optimized with High-Rez even should it otherwise be used to optimize the time-domain at mastering.
Regarding MQA comparison test proceedure, I would think that a common master test track file digitized from the analog source without any anti-alias filtering should first be produced by high oversampling rate ADC technique. Next, the relevant files targeting each tested format could be obtained by downsampling the oversampled master utilizing brickwall anti-alias filters for RBCD or High-Rez, and the MQA anti-alias filter for MQA. Then, compare the subjective sound from each format utilizing their respective downsampled versions of the common oversampled master test track file.
