Filler word dectection needs improvement

[Tom301685719t6l]

Filler word detection still lets a lot of uhms through, as a podcast editor i would still have to do a full manual sweep of the podcast to try and get rid of the rest of the uhms which doesnt really speed up my workflow

[dpalkowski3]

it's not flagging any filler words for me, just pauses

[Tom301685719t6l]

are you using the new beta version or normal?

[dpalkowski3]

yep, beta...i dont think the feature even exists in the release, no?

[dpalkowski3]

interesting: as an update. I exported the edited audio timeline as a single mono file and reimported it, now it seems to be detecting filler. Nothing was changed in the export, so maybe it doesn't like audio that has been imported as part of an mp4 video..

[Stan Jones]

@dpalkowski3,

From the staff post about this (link below), "If you’ve previously transcribed a clip or sequence, you’ll need to re-transcribe to add filler words to it." Is it possible the original file was transcribed before you updated to a version with the filler words feature active?

This tripped me up on the first times. Also, one test file I use a lot is voice over by a professional speaker, and had NO filler words.

Stan

[dpalkowski3]

aha, that's probably it then. I'd already transcribed it, then opened the project up in the beta version., thanks!

[dpalkowski3]

now, i hope they'll add a feature where you can create a list of filler words to be removed. It seems to do pretty well with uhh and umm, but what about 'you know', kind of', sort of', 'like', 'if i may', 'obviously', etc etc etc. Every speaker tends to have certain tics they use, so being able to specify a list of lookouts would be very helpful.  My current method using this toolset will be to use premiere to batch the umms, then export and open the file in Adobe Podcast to deal with the rest of the filler words (text editing is more responsive there, and it has the enhance feature already enabled)