| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a speech synthesis device and a speech synthesis program, capable of outputting speech without giving sense of incongruity. <P>SOLUTION: An interrogative sentence phonetic model 512 and an interrogative sentence prosodic model 522 are prepared, in advance, from the phonetic data of speech, in which an interrogative sentence, a sentence requesting agreement, a sentence prompting action, or the like whose tone of voice of the last part of the sentence is different from that of a declarative sentence are uttered. When the last part of the sentence is a prescribed pattern (when there is a question mark "?" attached, phrase for a prescribed question, phrase requesting agreement, or phrase prompting action), and when an interrogative sentence exists in the sentence, a natural speech that is closer to the tone of voice of the last part of the sentence of the interrogative sentence, the sentence requesting agreement, the sentence prompting action, or the like can be output, by selecting a sound model (phonetic model and prosodic model) from the interrogative sentence phonetic model 512 and the interrogative sentence prosodic model 522. <P>COPYRIGHT: (C)2007,JPO&INPIT |
