| 摘要 |
PROBLEM TO BE SOLVED: To provide an information processor capable of estimating more proper pronunciations by generating pronunciation rules for more properly estimating a pronunciation of a word being a processing object and using the pronunciation rules to estimate a pronunciation, and to provide a control method therefor and a program. SOLUTION: A representation character division part 102 acquires the word being the processing object, from a word dictionary 101 consisting of a plurality of words having information of representations and pronunciations and divides its representation into a plurality of partial character strings. A partial character string connection part 103 connects adjacent partial character strings out of the plurality of divided partial character strings to generate new partial character strings. A pronunciation rule generation part 104 determines a pronunciation corresponding to each obtained partial character string and registers a pair of the partial character string and the pronunciation in a pronunciation rule holding part 105 as a pronunciation rule. A pronunciation rule deletion part 106 deletes registered pronunciation rules on the basis of frequencies of pronunciation rules registered in the pronunciation rule holding part 105. COPYRIGHT: (C)2005,JPO&NCIPI
|
