EVALUATION OF THE RULE ACQUISITION ON A DIRECT SPEECH TRANSLATION METHOD WITH WAVEFORMS USING INDUCTIVE LEARNING FOR NOUNS AND NOUN PHRASES

摘要

This paper evaluates a direct speech translation method with waveforms using Inductive Learning Method for nouns and noun phrases extracted from real conversations. The proposed method treats only acoustic characteristics of speech waveforms of source and target languages without obtaining character strings from speech utterances. And our method is realized by learning translation rules that have acoustic correspondence between two languages recursively. Therefore, we are able to avoid some serious problems of conventional speech recognition and speech synthesis because syntactic expressions are not needed for translation. This speech translation method can be utilized for any language because the system has no processing dependent on an individual character of a specific language to realize speech translation. In this paper, we deal with a translation between Japanese and English.