Today, my home spoke to me. Twice. The first was to tell me to wake up, and the second was to tell my dinner was ready.

Do not adjust your set... this is not an episode of Lost... nor a dillusion. This is reality!

Using my Linux server and Festival, it is very easy to make the computer talk. A simple script like:

echo "(say \"$*\")" | festival --pipe

Allows me to invoke the speech by the usual home automation software. And by using at I was able to use the computer to remind me when to take the fish out of the oven. It's that simple.

But there is a snag.

It's scary! The synthesized phonemes, its unfeeling voice, and the unannounced manner in which the computer voice introduces itself is scary. Other worldy, even. There are several solutions I'm sure. One would be to have a woman record all the necessary phrases for me (people respond better to information from a female voice, but commands from a male, apparently) and piece those together. But audio Lego always sounds like train station PA systems, you're limited to the pre-recorded words, and I don't have the time to develop any better algorithms. Perhaps some good software already something similar with pre-recorded phonemes, but I'm after Free and Open Software.

Another version would be to adjust the default synth voice to something more soothing. Again, it's still a machine.

My current approach (and we'll see if it's scary when it announces something unexpectedly) is to introduce the voice with a jingle, before proceeding with the message. I'm also experimenting with the idea of always beginning with the name of the person to whom the message is addressed. And/or the time.