Speech synthesis & voice control

After spending some time thinking about the UI for WestiAmp, I've decided on a rough architecture:

1. A daemon (almost typed 'service' - my day-job as .NET developer is having an adverse effect on me) called westiplayerd, that checks for the presence of control files (empty regular files with particular filenames like 'nextplaylist' or 'stop') and plays MP3s / OGGs and playlists accordingly.

2. Another daemon, called westicontrold, that accepts input, either through voice (using a speech recognition program) or from the knobs on the panel (probably through a parallel port?) and touches control files which will be acted upon by westiplayerd.

3. A web application (probably written in PHP) which will allow a user to upload music files to the WestiAmp, and delete those files if needed. It will also provide a UI that displays what westiplayerd is up to (i.e. which track & playlist it's playing), and that will allow the user to issue commands to westiplayerd (again by touching control files).

I'm investigating KTTSD to add speech output to westiplayerd for the purpose of letting the user know about errors, full discs etc. As of yet I have no idea what speech recognition software I'll be using, but I do know there's quite a range available for Linux.