| 摘要 |
PURPOSE:To realize an intonation wherein the structure of an input sentence is reflected by gathering sentence structure indication vocabularies and generating a dictionary in advance, including information sentence structures in the dictionary, and generating an intonation matching with the structure of a recognized input sentence. CONSTITUTION:The dictionary is generated by gathering sentence structure indication vocabularies which indicate structures of sentences in a natural language, the information regarding the sentence structures are stored in the dictionary, and the intonation matching with the structure of the recognized input sentence is generated by using the information. Namely, the dictionary of those vocabularies contains the information regarding the sentence structures previously, so the information is utilized to recognize the structure of the input sentence right after morpheme analysis. Namely, a sufficiently natural intonation is obtained by structure recognition at a stage of vocabulary analysis. Consequently, an intonation is given to each group of structures and a pause corresponding to the intensity of a break is generated between groups of structures. |
