SPEECH SEGMENTATION SYSTEM, SPEECH SEGMENTATION METHOD, AND SPEECH SEGMENTATION PROGRAM

摘要

In a dialogue sound in which temporally overlapped speeches may occur, in order to segment speeches at appropriate positions to facilitate better understanding of a relationship between the speeches, a speech segmentation system comprises a speech segmentation means (120) for receiving speeches by two or more speakers and further segmenting each of the speeches. If another speech is made so as to overlap a certain speech, it is considered that a time point when the overlapping speech is spoken becomes a semantic segmentation. Therefore, the speech segmentation means (120), when determining a segmentation point of a certain speech, uses information on another speech that overlaps.