It turns out that it interfaces pretty cleanly to the Microsoft Windows text-to-speech facility, so that it takes less than six lines of code to be able to make the computer talk.
So, just before the holidays, one of the clowns in my unit at work was announcing his impending vacation (we all do, because many of us have specialties, and tickets often get left for those who have the relevant specialty - so letting others know about an impeding absence lets the 'backup guy' for the specialty 'ramp up' to cover the relevant tickets). Being the clown he is, though, he kept asking people "Are you gonna miss me?".
I had my "reply" all set up - I was going to make his computer suddenly say "Yes, I'll miss you. But I'll practice while you're gone, and my aim will improve.". There was a problem, though, which is clearly a bug in the MS TTS engine for US-English. It works fine for Canadian, British, Scottish, Irish, Australian, and New Zealand English, and with odd accents for Indian and Hong Kong English, but... you can't say "miss" with the US-English engine. At least with Windows 7. The US-English engine _invariably_ reads the four-character sequence 'miss' as though it were the eleven-character sequence 'mississippi'.
I have yet to try it on a Windows 10 box. I don't expect MS to be willing to bother fixing it for any earlier version of Windows.
ETA: I've now tried it with Windows 10, and they seem to have squashed the bug.