My hypothesis:
At low listening levels, the Fletcher-Munson equal-loudness curves (how our ears perceive bass and treble differently at quiet volumes) matter more than the type of speaker. Horns and high-sensitivity designs can sound more lively at low volume because they preserve dynamics and stay clear near the amp’s noise floor — but they don’t fix the ear’s natural loss of bass and top-end at low SPL.
If you apply proper loudness-compensation EQ, almost any well-engineered speaker can sound full and balanced at quiet levels.
Psychoacoustics dominates; equipment choices just help work around it.