With this logic, you should also be able to trim the parts that doesn’t have words. Just add a cut-off for db, and trim the video before transcription. Possibl

With this logic, you should also be able to trim the parts that doesn’t have words. Just add a cut-off for db, and trim the video before transcription. Possibly another 10-20% gain?