Lots of good comments above. As Bombaywalla and Larryi alluded to, I think the OP is conflating noise, and the ratio of signal-to-noise, with distortion in its various forms. Noise and distortion are very different things, having different causes and different effects.
As I see it distortion, which can exist in many forms, is what will typically cause listener fatigue, and also often result in the listener preferring to listen at a less than realistic volume and/or at a lower volume than he or she may otherwise prefer. And as Bombaywalla indicated, and assuming good source material, speakers and system components are likely to be the main contributors to distortion.
That is all consistent with comments Ralph/Atmasphere has made many times, to the effect that certain higher order odd harmonic distortion components are used by our hearing mechanisms as loudness cues, and minimizing or eliminating those distortion components is essential to the system being able to replicate, or at least approximate, the sense of "ease" which characterizes the dynamic peaks of unamplified music heard live.
Noise and the ratio of signal to noise, on the other hand, which I certainly agree can be helped in many cases by power conditioning, and in some cases also by vibration control, but which can also often be limited by the intrinsic performance of the electronics, involve low level effects that can adversely affect resolution, inner detail, ambience, micro-dynamics, and other such things. But I would not expect noise, or its relation to signal level (i.e., S/N ratio), to have much if any effect on the maximum volume that may be preferred.
Regarding the comments by Charles and Larryi about listening at relatively low volumes, I certainly agree that the ability of a system to perform well at those levels is a desirable attribute. But given the premise that we want our systems to reproduce well recorded music as realistically as possible, my feeling is that that cannot be accomplished without listening at a volume level approximating, or at least approaching, the volume level at which a particular piece of music would typically be heard when performed live. One reason for that being the
Fletcher-Munson Effect. The tonal balance of our hearing mechanisms is different at different volume levels.
Most of my listening is to classical music, and IME the dynamic peaks of most live classical music, when listened to from a decent seat in a good hall, are LOUD. Not only in the case of orchestral music, but even when it comes to small chamber ensembles, solo piano, etc. Jafant alluded to this point in his post earlier in the thread.
For that reason, while I almost always agree with just about everything Charles and Larryi have to say in these forums, with respect to this particular point, which as Larry indicated is subjective and a matter of individual preference, I see it somewhat differently.
Best regards,
-- Al