Method, system, and program for inserting caption

TL;DR: In this paper, the problem of determining a correct caption insertion position on the basis of the results of speech recognition processing of speech information included in video information and recognition results of the characters of image information contained in the video information is addressed.

Abstract: PROBLEM TO BE SOLVED: To decide a correct caption insertion position on the basis of the results of speech recognition processing of speech information included in video information and recognition results of the characters of image information included in the video information. SOLUTION: A frame image input means 2 inputs the image information while a sound input means inputs the speech information. A subtitle manuscript input means 4 inputs a caption manuscript including information which shows lines and speakers, while a caption manuscript storage means 5 stores the inputted manuscript. A speech recognition means 6 applies speech recognition processing to the speech information and detects the timing at which the lines are heard. A face recognition means 7 detects the timing at which the faces of the speakers appear in an image by relying on the image information. A caption insertion means 9 inputs information on the timing detected by the speech recognition means 6 and, after the information on the timing detected by the face recognition means 7 is inputted, the caption is inserted in the video information at a point of time corresponding to any of the timings. COPYRIGHT: (C)2005,JPO&NCIPI