摘要 |
PURPOSE:To accurately accent a synthesized word by controlling the rhythmical connection of adjacent words in a composite word and determining the position of an accent. CONSTITUTION:When a 'KANJI' (Chinese character)/'KANA' (Japanese syllabary) mixed sentence is inputted to a character code string input part 10 a morpheme analyzing part 11 divides a sentence into respective words constituting it and sends a part continuing nouns to be specific morphemes to a composite accent determining part 13. The determining part 13 refers a word accent connecting table 14, and in the case of connecting the accents of adjacent words, these word accents are connected by using an accent connecting rule. When the connection of accents is unnecessary, respective words are provided with independent accents. A synthesized sound parameter forming part 15 forms the continued time length of synthesizing words, the pattern of reference frequency and the parameter of a power pattern. A synthesized sound output part 17 outputs a synthesized sound waveform formed by a sound synthesizing part 16. |