TTS PROCESSING METHOD AND CONTINUOUS VOICE RECOGNIZING METHOD USING PRONUNCIATION-ACCESS GRAPH

摘要

PURPOSE: A TTS(Text-to-Speech) processing method and a continuous voice recognizing method using a pronunciation-access graph are provided to save a memory capacity and enhance overall performance used in the TTS system and continuous voice recognizing system by displaying a plurality of candidate pronunciations appeared in one morpheme as the pronunciation-access graph. CONSTITUTION: A computer readable recording medium for the TTS system comprises a word class field, a morpheme pronunciation field, a probability information field, a right phoneme access information field and a post pronunciation access node field. The morpheme pronunciation field stores the character row by which the morpheme is pronounced. The probability information field stores the probability in which the morpheme is pronounced in accordance with the character row stored in the morpheme pronunciation field. The right phoneme access information field stores the spelling and pronunciation of the last phoneme element of the morpheme. The post pronunciation access node field stores pointers for accessing to at least one pronunciation access nodes in which post morpheme pronounced after the morpheme is stored.