Since I was a child, I have been amazed by speech synthesizers. The first one which I could use was SAM, in the eighties, which was a program running on the Commodore 64. That kind of voices war very cold and robotic, but already understandable. The C64 audio chip (called SID) was very different form what you can find in toda’s hardware: it was an analog circuit with three "voices", or tracks.
Since then, hardware and speech synthesis algorithms have got better and better. The new "text-to-speech" programs use now big databases of professional speakers samples; the most used words and phrases are recorded, as well as all the phonemes which many languages use.
I have just found the following web site which hosts one of the most advanced and accurate speech synthesis so far:
http://www.ivosoftware.com/ivonaonline.php
You can let the voice speak in several languages. Excellent.