Auto-detect spoken double words in transcript

Similar to how it detects pauses, an Automatic way to detect (and delete/modify) double, double spoken words in the transcript would save almost as much time in correction workflow as the remove pauses did. I guess if it could find a very high match between words immediately beside each other? 

IT would be valuable to me because it would eliminate yet another thing I have to constantly edit out of presentation videos and would save a TON of time.