摘要 |
<p><P>PROBLEM TO BE SOLVED: To provide an utterance section identification apparatus for promptly detecting a speaker rotation place. <P>SOLUTION: The utterance section identification apparatus comprises: a voice data acquiring portion 20; a processing section determination portion 20; a processing section storing portion 30; a feature quantities extracting portion 40; a speaker rotation judgment portion 50; a speaker rotation judgment result storing portion 60; a result output portion 70; and an utterance section rule 80. A return point is set at time corresponding to two times of a minimum utterance time in which a single speaker speaks in a minimum period, from a speaker rotation place which is detected last time. Along with backward time direction from the return point, processing time for searching a speaker rotation place this time is determined, and along with forward time direction from the return point, processing time for searching a speaker rotation place next time is determined. <P>COPYRIGHT: (C)2008,JPO&INPIT</p> |