| 摘要 |
PROBLEM TO BE SOLVED: To synthesize voices nearer to natural language according to pause information by calculating relation between attributes of morphological units of a text and attributes of adjacent morphemes and processing the relation statistically to construct a pause decision rule highly accurately and to output the voices. SOLUTION: This natural language processing method divides a text into morphological units (S501, 502) and calculates a pause information decision rule stipulating the decision rule of pause information for every morpheme based on attributes of respective divided morphemes and attributes of adjacent morphemes (201, S503) and decides pauses of the whole of the text from pause information of respective morphemes by deciding pause information for every morpheme according to the decision rule and by imparting them to the text (S504). |
