This happens in the same way that users of a traditional computer word processor benefit from having more than one text font.
The Natural Voices software in the sources section allows defining acronym pronunciation and speech rate, as well as other things.
This may be similar to the difficulty in achieving true Artificial Intelligence that can actually pass a Turing Test by presenting spectators with something indistinguishable from what it is trying to simulate.
Like its text counterpart, each voice font can supply a different experience and provide a selection for different purposes.
The simplest one is to select a voice font from a group in order to get the clearest one, or to choose the one with a speed that is appropriate for different settings.