Automated voice translation dubbing for pre-recorded videos

摘要

A method is provided for aligning a translation of the original caption data with an audio portion of a video. The method includes identifying, by the processing device, original caption data for a video comprising a plurality of caption strings. The processing device identifies speech recognition data comprising a plurality of generated character strings and associated timing information for each generated character string. The processing device maps the plurality of caption strings to the plurality of generated strings using the assigned values indicating semantic similarities between the strings. The processing device allocates timing information to individual caption strings based on the timing information of the mapped individual generated strings. The processing device aligns the translation of the original caption data with the audio portion of the video using the assigned timing information of the individual caption strings.