in Ideas · 108 words

A tool for extracting subtitles from sub-in-picture fansubbed video. Use ffmpeg to decode the video, jocr for picking up the characters; have a GUI which allows the user to scroll around inside the file and pick a colour and range to match for subs (or several, in fact). Allow sub timing to be arranged easily (multiselect on a list).

The reason I wanted this was for transcribing the old Rurouni Kenshin fansubs to use with KentANIMEted's DVDs, but the quality is so poor that it's unlikely any OCR approach (let along jocr) would do a decent job. Transcribing the subtitles by hand takes about an hour an episode...