Strip “um,” “uh,” “you know,” and other filler from a transcript or caption file — right in your browser. Watch every cut, keep the ones you want, hold onto your timestamps, then copy or download the clean version.
Transcript
One per line or comma-separated. Phrases work too (e.g. "you know"). Matching ignores case.
Common but risky additions (can have real meaning):
Edit — struck words are cut · click one to keep it · type to fix the flow
What Copy gives you
Everything runs in your browser — nothing is uploaded or saved. Filler words are struck through in place: cut from the result, but still visible while you edit around them. Click any struck word to keep it (click again to re-cut), and type anywhere in the pane to fix the flow by hand. The lower pane is exactly what Copy and Download .txt give you. Import an .srt or .vtt and the caption timings stay out of sight while you edit — Download .srt/.vtt puts them back, with the filler removed.
Auto-generated transcripts and captions are littered with verbal filler — “um,” “uh,” “you know,” “I mean,” “sort of.” This tool strikes them out for you, but it never deletes blindly: every cut is shown in place so you stay in control. Keep a filler word where it actually belongs, fix the flow by hand where a removal reads clunky, then copy or download the result.
Paste your transcript, or import a .txt, .srt, or .vtt file.
Filler words are struck through automatically. Edit the text and click any word to keep it.
Copy the clean transcript, or download it as .txt — or .srt/.vtt with timestamps intact.
One recording becomes show notes, captions, a blog post, and clips. Each of those reads better without the verbal tics of natural speech. Clean the transcript once and reuse it everywhere — and because the timecodes in an .srt or .vtt survive the edit, your cleaned captions drop straight back onto the video.
Paste your transcript into the tool, or import a .txt, .srt, or .vtt file, and it instantly strikes through common filler words — um, uh, you know, I mean, sort of, kind of, and more. Each cut is shown in place so you can see exactly what's being removed; click any struck word to keep it. Then copy the cleaned text or download it as a file. Everything runs in your browser.
No. The transcript cleaner is 100% client-side. Your text and any file you import are processed entirely in your browser and are never sent to a server. There's no upload, no account, and nothing is stored — closing the tab erases everything.
Yes. Import an .srt or .vtt file and the tool hides the timecodes while you edit, showing only the spoken text. When you download the result as .srt or .vtt, each caption's original timestamp is reattached to its cleaned line, so your caption timing stays intact and the file is ready to re-upload to YouTube.
The safe defaults are pure verbal tics that almost never carry meaning: um, uh, er, ah, you know, I mean, sort of, kind of, basically, and literally. Words like like, just, so, well, right, actually, and really are riskier — they sometimes carry real meaning — so they're offered as an optional add-on rather than removed by default. You can edit the strike list to anything you want.
Only words on your strike list are touched, and every cut is shown struck through rather than deleted — so you can review each one and click to keep it before you copy. “Whole words only” is on by default, so striking um won't affect a word like summary.
It's completely free with no account, no signup, and no watermark. It's a free tool from YouPush, and you can use it as much as you like.
YouPush takes your finished long video and ships it to YouTube — title, thumbnail, tags, chapters, end card, and a week of vertical Shorts, in one pass. It doesn't clean your transcript; I built Strike as a free extra because it's useful. Local-first, bring-your-own-key, one-time purchase.
See what YouPush does →