We would initially need to calibrate the mechanism and microphones against expert violin players to ensure they pick up all the same nuances of sound. This is because experts hear more than non-experts.
"Two groups of expert instrumentalists (violinists and flutists) listened to matched musical excerpts played on the two instruments (J.S. Bach Partitas for solo violin and flute) while their cerebral hemodynamic responses were measured using fMRI. ... We found an extensive cerebral network of expertise, which implicates increased sensitivity to musical syntax (BA 44), timbre (auditory association cortex), and sound-motor interactions (precentral gyrus) when listening to music played on the instrument of expertise ..." http://cat.inist.fr/?aModele=afficheN&cpsidt=20951540
So if the experts could demonstrate any statistically significant ability to hear certain aspects of the sound, such as being able to distinguish between different instruments, this would have to be replicated by the mechanism and microphone technique. We would have to try to test a fair range of aspects of the sound.
We would then need a reasonably large sample of similar violins, of a quality for which we might expect the effect to occur, half of which would be put away and the other half played for our test period.
The mechanism could then be rerun and get a fairly good result.