How to Remove "Um" and "Uh" From a Video Without Paying Per Minute
Why speech-to-text hides filler words, how to find them anyway, and a free way to cut ums, uhs and repeated words from a video or podcast and keep control of every cut.
Filler words are the most common reason a good take feels amateur. Cutting them by hand means scrubbing the timeline for every "um", and the tools that do it automatically increasingly charge for it: Descript, for example, moved filler removal onto AI credits in its September 2025 pricing change. Here is how filler removal works and how to do it for free.
Why transcripts often don't show "um"
Speech models like Whisper are trained on cleaned-up transcripts, so they tend to leave fillers out of the text. That's nice for reading and bad for editing: a tool that only searches the transcript for "um" misses many of them.
Cuescribe's filler word remover does two things. It lists the fillers the model did write (um, uh, er, hmm), plus repeated words like "I I think" or "the, um, the door". Then it looks at the audio between words: when there is sound at speaking level but no word was written, that is usually a dropped filler, a breath or a laugh. Those are listed separately as "untranscribed sound" and are not cut unless you tick them.
A cut you can trust
- Review in context. Every item shows the words around it and plays from just before it.
- Preview the result. "Preview with cuts" plays the recording with every ticked cut skipped.
- Cuts land between sounds. Each edge moves to the quietest moment nearby, with a 5 ms crossfade, so there are no clicks or clipped consonants.
- Leave the tics alone. "Like" and "you know" are listed but not ticked: half the time they are real words.
What to download
The tightened video if you're done, a WAV for a podcast, an EDL if you want to keep editing in Premiere Pro, DaVinci Resolve or Avid, or subtitles already re-timed to the shorter cut. Long pauses can be shortened in the same pass.
Don't cut everything
A talk with zero fillers can sound robotic. Cut the ones that break a sentence, keep the ones that sound like thinking. That's why nothing is cut without you seeing it first.
Free, no upload. Get the extension for transcripts beside YouTube and TikTok videos, or see what Pro adds.