I think a lot of "harshness" might start as a problem of listener-perspective.
In recording, we're starting from the position of an unnatural observer - with mics postioned where ears would never be, and in (optimally) treated (or at least generally) in non-reverberant rooms.
In real experience we hear instruments along with reflections - in rooms, concert halls, whatever.
But recording has evolved to try to separate the instrument from its surroundings - so you get what, imo, is an "unnatural" isolated perspective when you listen back. Of course the character of the particular microphone adds to the filtering.
_
Lately I've recorded two acoustic instruments for the purpose of messing around in a sampler - piano accordion and 5 string banjo. Yeah, you can say those sound horrible to start with
But what I hear from the recordings is a harshness and some very unpleasant overtones.
I reckon the harshness and overtones are
really there ... just that in normal listening conditions they are softened by the surroundings; whereas in recording they are isolated and focused.
Maybe.
_