SPEECH SYNTHESIZING METHOD FOR A UNIT SELECTION-BASED TTS SPEECH SYNTHESIS SYSTEM

摘要

PURPOSE: A unit selection-based TTS speech synthesizing method is provided to generate synthesized speech with a small-size database. CONSTITUTION: A unit selection-based TTS speech synthesizing method includes a language processing step, a phrase generating step, and a signal synthesizing step. The language processing step pre-processes an electronic text, changes pronunciation transcription of characters that are exceptionally pronounced in the text with reference to an exceptional pronunciation dictionary, and analyzes sentences of the text using a sentence analysis grammar algorithm. The phrase generating step extracts phrases from the analyzed sentences of the text, sets a duration time of each of the extracted phrases, and sets intonation of each of the phrases. The signal synthesizing step selects a unit from candidate units, selects the unit from a unit database to synthesize a speech signal.