AUTOMATIC EXTENDING METHOD FOR JAPANESE TEXT

摘要

PURPOSE:To provide the automatic Japanese text extending method which can efficiently generate a text without newly generating a learning text. CONSTITUTION:In a Japanese text data base used for Japanese speech recognition, a statistical language model 2 regarding the appearance order of KANA (Japanese syllabary) and KANJI(Chinese character) is generated (step 10) as a statistical language model of a Japanese text data base 1, and the statistical language model 2 generated as the Japanese text data base 1 is used to automatically generate new text data 3 different from the Japanese text data base 1 (step 20).