I primarily use Descript to generate transcripts for legal transcription. There are many ASR options available, and I'm always trying out others to see which one gives the best output, balanced with the best price.

One of the reasons I continue to favor Descript is because of its Remove Filler Words feature. But I've long had a love/hate relationship with it because while it does remove a ton of filler words, it still left behind a ton of clutter, ignoring sometimes hundreds (in, say, an all-day trial) of stutters such as "I, I, I, I" or "the, the, the, the." That always made for tedious editing and an endless quest to find or create a macro or other tool outside of Descript to finish the job without taking forever or crashing.

After my latest transcription editing project, it occurred to me in hindsight that I didn't recall encountering the usual clutter of repetitive stutters. That gave me pause. So just to make sure it wasn't my imagination, I went through some of my old projects and found a short 7-minute project that still had the repetitive stutter text after Remove Filler words had been applied (old version). I re-uploaded the same audio to a new project, confirmed the stutter words were in the unedited original, ran Remove Filler Words again, and what do you know? All the stutter words were gone, with no clutter left behind!

Finally, I can give up the quest to find an external tool for finish this clean-up job. And I can enjoy the savings of time not having to clean it up manually while I'm proofing/editing the output.

Other ASR apps seem to lean toward one extreme of the editing spectrum or the otherwise They either do ZERO cleanup, leaving it all to the user, or they clean it up too much, removing all the okays and all rights and other things that must be left in with verbatim transcription. Descript has always done the best job of cleaning up what you want to get rid of - no more, no less. But with all the clutter it used to leave behind, it still left a lot to be desired. Now it's Goldilocks-perfect for my purposes.

Thank you, thank you, thank you, Descript Team! :)