Your analogy is bad (obligatory "and you should feel bad", but not really).
If speech were expressed with paint on canvas, cruel speech would be painting goatse or the like.
Harassment would be following someone around with your painting of goatse. Or any painting of anything they object to. It's the "following them around" part that makes it harassing.
The verbal analogue of throwing paint on someone would be yelling at them through a megaphone set at painfully high volume.
Every instance of speech is also an action and every action is also an instance of speech, and the distinction between a speech-act as speech and a speech-act as action is whether you're talking about the information content (the speech part) or the physical method of delivering that content (the action part).
Throwing paint or blaring painfully loudly through a megaphone are harmful actions, assault and battery in fact, regardless of the color of the paint you throw or the noises you make through the megaphone.
Following someone around and exposing them to images or sounds they don't like is harassment, regardless of the images or sounds; it's the following-them-around part that makes it harassing.
Images or sounds themselves, presented in a way that is not physically harmful to anyone (the way that loud sounds or a face full of paint would be), in a way that anyone can walk away from, are just speech, cannot harm anyone regardless of their content, and thus should not be regulated in any way regardless of their content.