Couple of quick comments:

#1 - if you have small speakers/drivers at medium high volume you are pushing the drivers to the point of non-linearity where cone movement doesn't track signal. With enough design effort you can make the speakers keep sounding good in that non-linear zone (Ian mentioned this explicitly) but if I remember correctly when you have multiple frequencies at the same time ("busy content") then you get IM distortion once the driver moves into the non-linearity zone

#2 - IIRC there isn't a sharp divide between "linear" and "non-linear", more of an "it starts getting worse at zero" kind of relationship

#3 - I have always preferred bookshelf speakers when I was playing at volumes they could handle - it may be as simple as a smaller driver array providing better imaging with all other things being equal - but for clean "wall of sound" louder playback I always end up with floorstanders

#4 - one other contributor to good bass on the M3 is the base bump between 70 and 150 Hz from cabinet/driver tuning (check the frequency response curve under Specifications). I find it enough to be noticed but rarely enough to be a negative

#5 - my impression is that cabinet resonances can also contribute to "busy" content sounding worse than "simple" content but I strongly doubt that is a factor on M3's

It's arguable, but I still find M3's to be some of the most surprising speakers you can listen to in terms of the perceived clarity of sound. I say "perceived" because strictly speaking the other Axiom models offer even a bit more clarity in the upper midrange and because a bit of the perceived bass clarity comes from the bass bump, but the overall result is very (very) "listenable".

The M5HPs are still "better" in pretty much every respect but it would be really interesting to run a varied group of listeners through an M3 vs M5HP test and see what they prefer. My guess would be at least 50% preferring M3's.


M60ti, VP180, QS8, M2ti, EP500, PC-Plus 20-39
M5HP, M40ti, Sierra-1
LFR1100 active, ADA1500-4 and -8