Descript's text-based editing is great for working on multitrack interviews, but if your audio tracks aren't fully isolated, your speaker's voices could be spread across some or all of the tracks causing speaker labels in the script to be assigned inaccurately.
Descript has the ability to detect microphone bleed, and gives you the option to automatically fix it so that your script stays clean, accurate, and editable.
How to automatically fix duplicate transcriptions
If Descript detects mic bleed in a multitrack sequence, you'll see a notification that says "Sections of (your_filename) have duplicate transcription, perhaps caused by mic bleed". Click Fix to remove the duplicate transcription.
Removes text, not audio
The Fix duplicate transcripts feature does not delete the analyzed media - it will only remove the duplicated text from the script.
How to manually fix duplicate transcripts
You can also manually initiate mic bleed detection on a sequence by opening the Sequences folder in the Project Sidebar, right-clicking the sequence you want to analyze, and selecting Advanced > Fix duplicate transcripts.
How to revert duplicate transcripts
In rare occasions, Descript can incorrectly categorize some audio as mic bleed. If that's the case, you can easily restore the script text by opening the Sequences folder in the Project Sidebar, right-clicking the sequence you want to analyze, and selecting Advanced > Revert duplicate transcripts fix.