METHOD FOR PREPARING LEXICON, METHOD FOR PREPARING TRAINING DATA FOR ACOUSTIC MODEL AND COMPUTER PROGRAM

摘要

PROBLEM TO BE SOLVED: To provide a method in which uttering is efficiently and accurately segmented into the words that are the constituting elements of the uttering. SOLUTION: A method of preparing a lexicon 32 used for preparing training data set for an acoustic model of Indonesian, includes the steps of: preparing a mapping data 54 between phonemes of Indonesian and phonemes of English; preparing a lexicon 56 including a plurality of the number of entries each including a word in Indonesian and phonemes in Indonesian associated with the word; and replacing the phonemes in Indonesian of each of the entries of the lexicon with phonemes in English to which the respective phonemes are mapped. COPYRIGHT: (C)2006,JPO&NCIPI