Given that lexicons are closed by their nature and that the input text for general text to speech (TTS) systems is open, there will always be words in the text which are not contained within even the largest lexicon. Even when a large lexicon can be constructed to cover the whole vocabulary it would be useful to find a principled method to reduce the size of the lexicon (which we discuss more fully in [11]).
In many languages the orthographic system has some relationship to the pronunciation, depending on the language it may be trivial (such as in languages like Spanish) or relatively difficult (like English), or harder (such as in Japanese with full kanji). Humans can (often) pronounce words reasonably even when they have never seen them before. It is that ability we wish to capture automatically in an LTS rule system.
Here we present a method for taking large lists of words and pronunciations and building generalized rule systems that not only produce reasonable pronunciations for unseen words but also allow us to remove the regular examples from the list so that much smaller lexicons are adequate for the same coverage.