Re: [Accessibility] Why not first an IDE that recognizes speech?

[Eric S. Johansson]

 On 7/28/2010 12:35 PM, Susan Jolly wrote:
> I'm a Tab.  Capital T to remind me to emphasize that at my age temporary is
> very likely to be true.  I'm a retired computational scientist (numerical
> solution of coupled nonlinear PDEs, etc.).  I saw the announcement for this
> list as a result of my interest in braille software.
>
> Never knew much about speech recognition before but the posts that point out
> that it is REALLY HARD have convinced me.  I know really hard from an
> algorithmic standpoint.  Sounds like a bad first project.
>
> My impression is that a lot of posters here really want to be able to use
> speech recognition to do software development.  Why not start with that?
> Interesting important problem with IMHO a greater chance of success and
> might well support the harder project both as a developer tool and as an
> extensible code base.

yeah, it's a really good idea but this is where the conflict comes in. Honoring 
Richard's philosophy of doing nothing that encourages or leads users to use 
proprietary software, we can't do thiss. NaturallySpeaking is the only useful 
system out there (ssee my other post about programming issues with speech 
recognition) and in order to do programming by voice right, you need to deeply 
couple the speech recognition application to the programming environment. And we 
are back in conflict began.

Haven't found a way to solve that problem yet. Not sure it's even possible short 
of forking the work into tools and application recognizer bridges as a 
non-canoe[sic] project and leaving a minimal component that can be built using 
the free software foundation philosophy in mind to them.

I need to go back and reread some of the messages but I believe (and please 
correct me if I'm wrong because this could help solve part of the problem) that 
the free software foundation philosophy not allow us to build all of the 
toolkits and bridge software if the only recognizer available was 
NaturallySpeaking. Of course, one way to solve that might be to put a Sphinx 
recognizer in place and let it exist as a time wasting honey trap until the real 
free software foundation recognizer was working.  that way, the free softtware 
foundation philosophy is satisfied (completely free toolchain) and those of us 
who know better... I mean who need to get  work done...er... are totally evil 
could use NaturallySpeaking as a part of a non-free software foundation 
component that works with the free software foundation tool chain.

My snarky humor aside, would that work? Would using Sphinx as a placeholder 
satisfy everyone for the short-term as we build for the future?