| 摘要 |
PROBLEM TO BE SOLVED: To provide a voice translation device, a method and a program of voice translation capable of recognizing voices of a pronunciation which is mostly used in a conversation.SOLUTION: The voice translation device includes: a voice recognition section that recognizes and converts a voice of a first language into an original sentence text while referring to voice conversion dictionary; a translation section that translates the original sentence text of the first language into a translation text of a second language; an unknown word detection section that, in the translation text of the second language translated by the translation section, detects an unregistered language in a voice recognition dictionary for recognizing a voice of the second language as an unknown word; a pronunciation estimation section that estimates a first pronunciation candidate for the pronunciation of the unknown word detected by the unknown word detection section based on a notation of the unknown word in the second language, and estimates a second pronunciation candidate of the unknown word based on the pronunciation of the word of the source language in the first language; and a dictionary update section that registers the unknown word and the estimated first and second pronunciation candidates in association with each other in the voice recognition dictionary. |
