[camlorn]

I think that the major shift to graphical interfaces with muds is more to do with laziness and making them accessible to a wider audience than it is with making competition easier. Obviously people mudded before the invention of graphics or the miniwindow. Multiple voices probably wouldn't help.
The top rate achievable by Espeak is, according to the manual, 900 words a minute. This is subject to change depending on the passage: a word is not like a meter or a foot. It is possible to flush the speech buffer, and I and many others do that; this is close but not equal to skimming, and allows one to start skipping combat messages. Moving the prompt to a hotkey is something that I've done, and there are various auditory progress bars one could implement. Alter Aeon has these via MushZ, but I've not felt it worth my time to bother implementing one for other muds, even though such would take all of a half hour, at most, and would probably be reusable by simply modifying one trigger. It is possible to skim up the input, as it were, but it isn't easily possible to skim new stuff coming in.
I haven't looked at the piece of software you've found but suspect that it wouldn't work well; taking control from the person to whom the reader is reading would probably add to the confusion. The main method of skimming is to read the first few words of each paragraph/line/etc, but the formatting information is lost when mudding; events are separated by time, not lines, for us. When reviewing prior output the formatting information exists again, just not while reading new messages.