| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a translation accelerating method or the like which can be applied even to a device with a low throughput. <P>SOLUTION: An automatic speech translation device 1 automatically translates voices by a storing part 30 for storing a translation library 31 in which words or phrases are grouped and registered in each different subject and a standard translation time 32, a speech recognition means 21 for recognizing input speech information to generate text, a translating means 22 for searching the translation library according to a search priority order set in a group and translating the text, and a speech combining means 23 for converting translation text into translation speech information. A search priority sequence changing means 24 measures a translation time required for translation by the translating means a plurality of times, compares a plurality of translation times with the standard translation time, determines the subject of input voice when the comparison results satisfies a predetermined condition, and changes a search priority order of a group corresponding to the determination results to a rising direction. <P>COPYRIGHT: (C)2008,JPO&INPIT |
