1) Buy an expensive HATS, don’t know how much a GRAS is, but the Neumann one is >$8000.
2) Record all speakers in the same position, same average SPL, playing the same tracks.
3) Create a DBT.
However, the sound system you are listening on needs to be as accurate in terms of tonal balance, imaging, soundstage, etc. as our ears, so it would all be for naught. Maybe include the actual song as a reference (maybe do an ABX, with a speaker and the actual song, the one that is closest to 50/50 is the most realistic speaker).