[as-devel] Voice Recognition and Afterstep

Derek Wyatt (derek@scar.utoronto.ca)
Thu, 1 Jul 1999 13:22:39 -0400 (EDT)



-- I was asked to send this to this list, so......

I have been playing around with ViaVoice from IBM recently.  It isn't
ready for a dictation type of scheme right now, especially considering
that it can't be trained for a particular person's voice at all easily.

However, it handles simple grammars very well!  Something like "window
resize twenty five eighty" is detected with about 95% accuracy.  ones that
don't come close enough are simply rejected.  It's extremely rare that it
causes some 'strange' behaviour.

I am going to integrate voice recognition into version 1.7.111 for my
own purposes.  Now, i asked a few people in #afterstep, and they thought
that it might be interesting to have.  So i figured that it might be a
good idea to integrate this into AS 'nicely' so that the feature could be
enabled on configure if one wanted it.

Do you agree that this may be a good idea?

If so, what would be the best way to integrate the code?  I haven't looked
at the source for AS yet, so that's going to be something i look at today.

Would this best be done as a module (i hope!) or does it need to be
integrated into src/afterstep/* ?

Cheers,

Derek

-----------------------------------------------------------------------
Derek Quinn Wyatt        | Christian : I will pray for you.
Unix Admin / Programmer  | Atheist   : Then I will think for the
derek@scar.utoronto.ca   |             both of us.
-----------------------------------------------------------------------