I've got some scripts for controlling mpd with the device in my pocket (via powerlaunch). And another script, that I can't post now.

And, used by all of these, the 'say' script. It's just a wrapper around flite, sox, and esdcat. It plays out of only one channel, and at variable pitch/speed and volume. Usage below.

I put all these scripts in /home/user/apps, so the absolute pathnames may need edited... but if you're looking here, you probably can handle that.

• rate is the playback sample rate. 8000 plays the voice as it comes from flite. Lower numbers give a lower pitch, and take longer; higher values give a higher, faster, more robotic-chipmunk-sounding result.
• vol is the volume scale factor. 1 gives same as flite, beyond 10 only increases distortion without boosting volume substantially. Values from 1 - 5 are helpful; after 5 the distortion starts getting annoying.
• text is just the text to speak. Since we're practicing script fu, that's usually a backquoted pipeline.

Dependencies:

• flite is in some repo, extras I think? Installs from Package Manager.
• sox is in some repo -- Gronmayer.com it if you need to. I think it was a Bora repo, and I think you need to redpill or apt-get install it manually.
• esdcat was by far the hardest bit to find. Sid Lenny armel to the rescue -- I installed both esound-clients and esound-common, then copied /etc/esd.conf (original OS2008 file) over /etc/esound/esd/conf (new file from esound-common package). It might work by simply forcing a dependency override, but IMHO better not to go that way... Anyway, when you're done, you should have esdplay and esdcat, which should have been available out-of-box, I would think.

Those are all attached. Here's a lame one, for VNC over SSH. This one's small enough to in-line:
And finally, a one-liner for launching/unminimizing xmms. Just put this line directly into osso-statusbar-cpu.
Edit: I should mention that the mpc* scripts use nc (aka netcat), which is to be found in the repos as well.