Speech synthesis, how to achieve perfect sound?

Question

speech synthesis

Here is my example: yadi.sk Acapela Alyona SAPI-5 Balabolka.
And here is an example of RSS-Radio: rssradio.ru What kind of voices do they have, a speech engine? How to achieve the same?
Who is fond of this, tell me how you set it up? Submit your samples. Where did they get dictionaries, homographs, etc. Or maybe you made it yourself, share links.
I, for one, use it to sound books, and you?
UPD: Found voices with perfect sound: https://www.ivona.com/us/