Desktop speech configuration

[Matthew Garrett <mjg59-gal srcf ucam org>]

We currently have two main reasons to use speech on the Gnome desktop:

1) To provide information to those who would otherwise be unable to 
obtain that information - probably the preserve of Gnopernicus in screen 
reader mode at present
2) To fill in for people who are otherwise unable to speak - Dasher 
implements this, I'm not sure about anything else

The two have quite different needs - (1) will be heard by someone who is 
going to hear it a lot, and so optimising for speed of information 
presentation is sensible. (2) is likely to be heard by a wide range of 
people, and so should probably be as close to "normal" human speech as 
possible.

Currently configuration for (1) is dealt with entirely within
Gnopernicus - presumably the idea is that someone who falls within the
catagories that require (1) will be using Gnopernicus for all of their
speech needs. I'm not sure that this is necessarily the case for (2) - 
we can certainly imagine that a user may wish to use different means for 
producing text for different purposes, but may want all of it spoken. 

What's the best way of resolving this? It seems to me that there ought 
to be a system-wide speech configuration system which could perhaps 
offer defaults of (1) or (2) - the number of people requiring both would 
seem likely to be small (perhaps applications should be able to flag the 
purpose of their speech output?). Having to configure speech in multiple 
applications seems like the wrong way of solving the problem, even if 
most users would tend to only require speech from a single application, 
but currently it seems to be the only way of doing things.

-- 
Matthew Garrett | mjg59 srcf ucam org