Text-to-speech synthesis (TTS) was at one time more commonly called synthesis-by-rule. It is a process which takes text in ordinary spelling and produces spoken (synthesized) output. There is particular interest in this in countries where the spelling system is not a good guide to pronunciation; good examples are English, and the Kanji writing system used for Japanese.
The best description of how text-to-speech synthesis works is in J. Holmes and W. Holmes Speech Synthesis and Recognition, Second Edition, 2001, Taylor and Francis, Chapter 7.
[Introductory material to be inserted here]
A number of companies and research groups who work on TTS provide demonstrations on their web-sites. You can try the following: