Remove Filler Words From Video Free (Ums and Uhs)

There are two different jobs hiding inside "remove filler words from video," and it's worth being straight about which one this is. Cutting the ums out of your video's audio is an edit. Keeping them out of your captions, so every stumble isn't immortalized in bold animated text, is one free click in ClipCraft's Video Captions tool. This post covers that click in detail, and then deals honestly with the audio question, because a caption cleanup that pretends to be an audio cleanup would be a scam.

One button that counts before it cuts

The 🧹 Remove ums & uhs button sits in the captions editor, and its label carries a live count of what it would strip; when I pasted a deliberately clumsy 60-word take, it read "(5)" before I committed to anything, and it grays out at zero so you never wonder whether it did something. It works on a transcript you've typed or pasted, and it works just as well on captions that already exist, including ones the AI auto-captioner generated from your video's audio.

Which words it removes, and which it deliberately doesn't

The ⚙ Which words panel has six toggle groups. Ums and uhs are on by default, with matching loose enough that "ummmm," "uhhh" and "mmhmm" all count. The word groups ("like," "you know," "I mean," "kind of / sort of," "basically / literally") are off until you turn them on, and that default is the feature's best judgment call: a real "like" is often doing actual work in a sentence. On my test take it stripped every um and uh and left "you know" and "basically" untouched, which is exactly what I'd have chosen reviewing it by hand. A custom field takes your own comma-separated words and phrases ("dude, at the end of the day"), and the whole configuration is remembered in your browser for next time.

The details that keep captions readable

• Punctuation gets rebuilt rather than left stranded: "I was, um, going" comes back as "I was going," with the orphaned commas cleaned up.
• On timed captions, per-word highlight timings are re-indexed to the kept words, and if a line started with a filler, its start time shifts to the first real word.
• A caption that was nothing but filler gets dropped entirely, and its screen time is handed to the neighboring line so the captions never blink off mid-sentence.
• It's free: client-side, no tokens, no cap on uses.

What about the audio?

The um is still in your voice track; the captions just stop transcribing it. For a lot of content that's the right amount of fixing, because viewers forgive a spoken stumble and screenshot a captioned one. If the audio itself needs work, the honest toolbox today is: our Silence Remover for the pauses (which usually flank the ums anyway, so trimming them tightens the take), and a manual cut in the VideoCraft timeline for an um you truly can't live with. A word-level audio deletion tool, Descript-style, is not something ClipCraft has today, and I'd rather tell you that than let a headline imply it.

The workflow around the button

Captions get into the editor three ways: paste a transcript (free, with tap-to-sync timing), let ✨ Auto captions transcribe the video with word-level timestamps for 5 tokens per started minute, or type lines by hand. Then the filler pass, then a style from the gallery (82 and counting; the TikTok captions guide tours them), then a burn-in export rendered in your browser. For Shorts specifically, the vertical-format walkthrough covers safe zones and sizing. The cleanup button itself never costs tokens, so the free path (paste, clean, sync, export) is genuinely free with a free account, and the token math for the AI transcription is on the pricing page.

Um-free captions in one click, your own banned-word list, and no charge for any of it. Make the free account, paste your last video's transcript, and see what the counter says about you.