Talking computers are generally candidates for ballistics testing, but my friend James had a neat idea for a debugging assist: tail -f application logs looking for certain patterns, then use Win32::OLE to play .wav files. Match /null pointer exception/i and the box goes BING!
This trick can be generalized by installing the Microsoft Speech SDK, which provides an OLE object, SAPI.spVoice, which can be told to speak arbitrary text.
So now, instead of having to switch windows to poke through logfiles, a semi-pleasant female robotic voice instantly informs me of my shortcomings as a programmer.
(Updated to note that I'm working out of earshot of others.)
For a monitoring tool
Using it for production servers is fine. Then some managerial person saw it, went "oh, cool, a talking computer monitor thingiee" and insisted everyone use it.
What's the chorus of "Server X going down" sound like (when the tool now monitors development servers ) ?. Noise pollution
