作者: Daisuke Nishiwaki
DOI:
关键词: Speech recognition 、 Segmentation 、 Character (mathematics) 、 Word (computer architecture) 、 Character recognition 、 Psychology 、 Image (mathematics)
摘要: There are provided: an n-fold-character recognizing part for collectively unmatched portion without segmenting character candidate patterns by image of a read-wise skipped portion, i.e., the upon word verification; and dictionary referred to recognition; thereby conduct re-recognition independent instability segmentation even when verification includes two or more characters.