Speech Synthesis from Unrestricted Text Using a Small Dictionary

摘要

Present voice synthesis systems have either used a large dictionary to look up words to be voiced, or have used only letter-to-sound rules resulting in monotone speech. The system outlined in this thesis, SIMON, utilizes the morphographemic analysis aspect of the large systems, while relying on letter-to-sound rules for the translation of morpheme to morph. The result is a system which produces inflected speech without the storage requirements of a large dictionary. (Author)