摘要 |
<p>PROBLEM TO BE SOLVED: To perform smooth conversation. SOLUTION: A voice recognizing part 1 performs voice recognition of voice inputted there and supplies the voice recognition results to a machine translating part 2. The part 2 applies machine translation to the voice recognition results from the part 1. When a user in Japanese has a conversation with a user in English, a conversation managing part 5 stores the conversion history and decides, e.g. whether translated results by the part 2 are natural, etc., from the flow of conversation on the basis of the conversation history and preliminarily stored knowledge. When the translated results are unnatural, the part 5 outputs the effect that the translated results are unnatural to the part 2, and the part 2 retranslates the translated results.</p> |