I don't know either, but it looks like the transcripts are initially generated from speech recognition (text dump + timing meta data?) and then hand-edited/annotated by a producer.
They'd add punctuation, sound cues, fix spelling, annotate the speakers (e.g. name + host, subject, or interviewer). Then that data's got to go somewhere...
It looks pretty labor intensive. I sure hope they have great tools!
They'd add punctuation, sound cues, fix spelling, annotate the speakers (e.g. name + host, subject, or interviewer). Then that data's got to go somewhere...
It looks pretty labor intensive. I sure hope they have great tools!