| 摘要 |
PURPOSE:To obtain highly recognizable synthesized voice by setting the style of a text, dividing the inputted text into morphemes and adding the way of reading of the text devided into morphemes and the accents in accordance with the set style. CONSTITUTION:A sentence mixed with KANJI (Chinese characters) and KANA (Japanese alphabets) are inputted from a character column input terminal 0. The sentence is divided into morphemes by a morpheme processing section 1a of a language processing section 1 employing a dictionary 1b and the reading of each morphemeaccent type and parts of speech are added. Then, a sentence structure analysis section 1c divides the sentence into sentence elements employing the information of each morpheme obtained by the section 1a and analyzes the relationships between the sentence elements. On the other hand, a sentence style setting section 1e sets the style of the inputted sentence, a reading/accent processing section 1d decides the accent phrase, the accent position, the pause, the intonation reestablishing position and the reading in accordance with the set sentence style employing the sentence structure analysis result. |
