3 ms·I've done this with Google's video transcription model, which works surprisingly well.by lights0123 4y agoI've done this with Google's video transcription model, which works surprisingly well.