We're looking into ways to improve the process of correcting machine generated transcripts while maintaining the timings.
Particular focus is given to ease, efficency, speed and accuracy.
Google Doc