transcription process

1 - full transcription and karaoke

write whole text
play media and press key at every word start

2 - captcha chunks and reduce

divide video into smaller chunks
serve chunks to several people, store different sub-texts
reduce submitted proposals into highest ranked words

3 - speech recognition, human validation

find and use good SR engine for PT, obtain words with timing
replay transcription, allowing word change (CRUD) and respective time shift

4 - hybrid, from subtitles

parse subtitles. use text and fill time-untagged words with interpolation between time-tagged words
replay transcription, allowing word change (CRUD) and respective time shift