I'm tempted to believe these results, as I suspect a lot of the reputation of stradivarius violins is overblown. However, there is almost always a huge problem

I'm tempted to believe these results, as I suspect a lot of the reputation of stradivarius violins is overblown. However, there is almost always a huge problem with these sorts of tests: the pepsi challenge problem. With subjective measurements it can often be difficult to robustly test the differences between two things in a short sample period. People may say one or another sounds better but that might only apply for a small sample. Or, people may just be unable to explicitly say which one they actually prefer when confronted with the choice in the moment. Whereas with regular use over extended periods of time they may be able to develop a stronger preference and be able to pick out differences explaining that preference. As a sort of proof of concept, aurally, I've noticed that there are several popular songs which have very distinctive audio glitches in them (typically high pitched beeps). The thing is, these glitches are not noticeable by most people when casually listening (testified by the fact that the songs are out in public release with such glitches) but once you notice them you begin to notice them every time, and it detracts from the enjoyment of the song. Similar effects on a subtler level surely exist for almost all music, so I wouldn't take the idea that listening to two separate samples of music and being forced to express a preference in the moment is at all a reasonable measurement except in more pronounced cases.