@Sightless combat, While I absolutely agree with you that this is an area where there will be varying opinions, One thing I will say is to not confuse inequities in design with the method of delivery.
Inability to interrupt speech isn't unique to screen reader output at all, up until recently a lot of interpreters had this problem when using Sapi output.
Obviously any system of game output needs to responsive to what the user wants, but that responsiveness and efficiency isn't specific to the discussion of the actual sound and output itself.
For me screen readers haven't quite got to the point where I can get all nuances from text, especially with dialogue that I can with a human narrator.
The vocal rhythm word pauses and intonation quality just hasn't been worked out efficiently, though it improving all the time, quite aside from all the of actual showing emotions and voice colouring that a good performer can bring to any piece of text.
This is why, while I do plenty of work with screen readers and even play gamebooks, muds and other interactive titles with them, when reading for pleasure, or heck even when reading for my thesis which requires some hideously complex articles on ethics with long winded language, I will always choose a human narrator if the option is there.
Of course I use screen readers a great deal, and the ease of production, as well as customizability have their place too.
If I could choose the optimum I think I'd like to see games things like menus and numerical statss are read by screen reader, but the discourses and in game dialogues arer voice acted, ---- just like in A hero's call in fact .
With our dreaming and singing, Ceaseless and sorrowless we! The glory about us clinging Of the glorious futures we see,
Our souls with high music ringing; O men! It must ever be
That we dwell in our dreaming and singing, A little apart from ye. (Arthur O'Shaughnessy 1873.)