I'll make this observation. Quick thought experiment.
Say you only have a single flash drive plugged into the BDP. Nothing else. It has the same track in flac and wav. Let's assume they sound different and we are able to double blind this. It holds up statistically. The hardware itself has not changed. Only thing that changes is FLAC and WAV. At this stage, if you had to guess, do you think the ultimate cause of this was in the software or hardware domain? I'm sure there will be ripple effects and proximate causes as well, which further muddy up things. But think of the ultimate cause.
Regarding WAV and FLAC, I've read most, if not all, of the previous Bryston threads on FLAC and WAV. Some hear it, others don't. I'd say most don't. Furthermore, most are not as likely to even entertain the difference existing in this case.
Let's switch over to Roon. Both MPD and Roon are bit perfect. Let's have the same devices, switches, and cables in play. Nothing else plugged into the BDP. The computer will act as both Roon Core and NAS, for both Roon Ready and MPD respectively. This ensures that the hardware used and the transmission method (ethernet) will be the same between Roon and MPD.
With Roon and MPD, a lot more people come out and notice a difference, even if you control for MPD being used through ethernet or local USB drives. This is not exclusive to Bryston. Go read the Roon forums and you'll see numerous discussions from customers of other manufacturers voicing the same thing. Different sound between local processing vs streamed music, even if the music is delivered by ethernet in both cases.
In this second scenario, let's assume we did a double blind test and found the differences to be statistically significant. Keep in mind that everything was same hardware wise. Would you think the ultimate cause for this was software generated. I'm sure the proximate causes for this will be even more interesting.
Keep in mind that people find the difference between MPD and Roon to be bigger than FLAC vs WAV. If the difference between MPD and Roon is software generated and this is something that people seem to easily accept, then why couldn't this be applied and extended to FLAC vs. WAV?
To make sure I don't botch this:
1) Hardware is kept constant.
2) Bit streams are perfect in all mode and mathematically identical.
3) Any sound difference generated is the ultimate result of how the software differs. This difference in software is potentially resulting in different amounts of generated noise (CPU and I/O). It's the hardware that ultimately produces the noise and generates any differences in sound. However, it's the software that governs how the hardware behaves.
Another important distinction that needs to be addressed in FLAC vs WAV. People will often say that the differences in decoding and handling is minimal and should not be audible. With Roon and MPD, I've seen people here more willingly to believe that this processing difference is enough to cause a difference in sound.
--> If we assume this to be true, then the next challenging question becomes: At what point does the difference in software and system resource usage stop being audible?
I'll leave it here for now.
EDIT: I will also add that we all use the term "noise" very loosely, especially when talking about CPU generated noise. We don't talk about the bandwidth, pattern, intensity, frequency whatsoever.