New: multi-track editing. Layer B-roll or add your own music on top of your video, free. Works on your projects, new and saved. Try it free in the editor, and send bug reports or feature requests to hello@vidclean.net

Blog

How to Remove Filler Words From a Video for Free

By Melvin Bucio 6 min read

You do not hear your own filler words while you record. You hear all of them on playback. Every "um" while you found the next sentence, every "uh" bridging a thought, every "er" that crept in because you did not want to leave a gap.

Cutting them by hand is miserable work. A filler is a fraction of a second, there are dozens of them in a ten minute recording, and each one needs a zoom, a scrub, and two cuts on the timeline.

Every tool worth using solves this the same way: transcribe the audio first, so every word carries a start and end timestamp, then cut at those timestamps. What actually differs between tools is what it costs, whether you need an account or a desktop install, and whether you get to review the cuts before they happen.

This post covers four methods. The first is free, runs in a browser, and exports MP4.

Method 1: VidClean Editor (Free, No Account)

Free filler removal on video, in the browser, with an MP4 at the end.

How it works. Open the VidClean editor and drop in your file. It transcribes the audio and gives every spoken word a precise start and end. Each um, uh, er, ah, and hmm is highlighted in the transcript. Press "Remove filler words" and the editor tells you exactly how many it found and how much total time they account for before it cuts anything. Confirm, and it cuts at those timestamps and rejoins the video. Export MP4.

You see the list first. This is the part most tools skip. The count and the total duration appear before the cut, so you can tell the difference between "twelve fillers, four seconds" and "two hundred fillers, ninety seconds" and decide whether you want all of them gone.

What it costs. Nothing. Transcription, filler removal, and MP4 export are free with no account and no watermark, for source files up to 90 minutes and 2GB. Pro extends that to 3 hours and 5GB.

Why it does not sound choppy. Each cut is pulled inward by 40 milliseconds so the words either side keep their edges, and every seam gets an 8 millisecond fade so the join makes no click. Back-to-back fillers like "um, uh" merge into one cut instead of two, which is what stops a dense patch of hesitation turning into a stutter.

Audio only? If you want a cleaned MP3 rather than a video, there is a dedicated filler word tool. The flagged transcript there is free; the cleaned MP3 download is a Pro feature. For video, the editor is both the better fit and the free path.

Limitation. Filler removal only cuts filler sounds. It will not tighten your pauses. If your recording also has long gaps, run silence removal as a second pass, or do both in the editor at once.

Method 2: CapCut (Desktop App, Account Required)

The most searched-for answer to this question, and the one with the least stable story.

How it works. In CapCut Desktop, add your clip to the timeline and generate captions for it. Filler word removal lives in the text-based editing panel that appears alongside the generated captions, as an option to strip hesitations from the transcript and the timeline together. There is a mobile path as well, but the desktop workflow is the one most tutorials cover.

What it costs. This is genuinely unclear, and it is worth being straight about that. CapCut's own tool page for the feature says you can remove filler words "for free with CapCut's trial". A lot of third-party coverage describes it as a CapCut Pro feature. There is also a CapCut community thread from users whose "Remove filler words" option disappeared from their Pro install. What is gated in CapCut moves around between versions, so check your own build before you plan a workflow around it.

The catch. You need an account, and for the reliable version of this feature you need the desktop app installed. You also inherit CapCut's terms around uploaded content, which is a consideration for some people and irrelevant to others.

Best for. People already editing in CapCut who want to stay in one tool. If CapCut is not already part of your workflow, installing it to remove ums is a lot of setup for one task.

Method 3: Premiere Pro (Creative Cloud Subscription)

The best version of this feature if you already pay for Creative Cloud.

How it works. Transcribe your sequence, then open the Transcript panel that powers Text-Based Editing. Click the filter icon and choose the filler words option. Premiere highlights hesitations such as "um", "ah", "uh", and phrases like "you know" directly in the transcript. Delete them from the transcript and they come out of the sequence. Because it works off Speech to Text, it is language agnostic and covers the same languages Premiere transcribes.

What it costs. It is part of Premiere Pro, so it costs whatever your Creative Cloud plan costs. There is no free tier.

The catch. Detection is not perfect, which Adobe has said plainly since the feature was in beta. Expect to review the highlights rather than trust them blindly. And this is a heavyweight NLE, so it is only a sensible answer if you are already editing in it.

Best for. Editors already working in Premiere who want filler removal in the same pass as the rest of the cut.

Method 4: Descript (Free Plan, One-Time AI Credits)

The tool that popularised transcript-based editing. Read the free plan carefully.

How it works. Upload your file, Descript transcribes it, and "Remove Filler Words" strips hesitations out of the transcript and the media together. As a workflow it is excellent, and it is the reason most people know this category exists at all.

What it costs. The free plan gives you 60 minutes of media a month, exports at 720p, and puts a watermark on video exports. Remove Filler Words is listed as limited on that plan and runs on AI credits, of which the free tier gets 100 as a one-time allocation rather than a monthly refill. So the free plan will let you try filler removal, but it is not something you can keep doing on it every week.

Best for. People who need a whole production workflow, not just filler removal, and are ready to pay for it. If removing ums is the entire job, this is far more tool than the job needs.

Which Method Should You Use?

It comes down to what you already have installed and what you are willing to pay.

You want the ums gone from a video, free, today: Use the VidClean editor. No account, no install, no watermark, MP4 out, and you see the filler count before anything is cut.

You already edit in CapCut: Use CapCut, but confirm the option is present and unlocked in your version first.

You already pay for Creative Cloud: Use Premiere's Text-Based Editing filter. Review the highlights rather than trusting them.

You want one tool for transcript editing, clips, and collaboration: Use Descript, and price the paid tier, because the free plan's one-time credits will not carry a weekly workflow.

For most people the honest answer is that filler removal is a small job that should not require a subscription or a download. Upload the recording, look at what was flagged, cut it, export. Then take the tightened file into whatever editor you were going to use anyway.

Frequently Asked Questions

What counts as a filler word? expand_more

Common spoken hesitations: um, uh, er, mm, ah, and hmm. Some tools also flag discourse fillers like "you know" and "like". VidClean matches the hesitation sounds using word-level timestamps from the transcript, then rejoins the audio so the result still sounds natural.

Can I remove filler words from a video for free without an account? expand_more

Yes. In the VidClean editor, transcription, filler word removal, and MP4 export are free with no account and no watermark, for source files up to 90 minutes and 2GB. Pro extends that to 3 hours and 5GB.

How do you remove filler words in CapCut? expand_more

In CapCut Desktop, generate captions for your clip, then open the text-based editing panel and enable the option to remove filler words. CapCut's own tool page describes the feature as available with its trial, while a lot of third-party coverage describes it as a Pro feature, and some users have reported the option disappearing between versions. Check what is gated in your own install before you build a workflow around it.

How do you remove filler words in Premiere Pro? expand_more

Transcribe your sequence, open the Transcript panel used for Text-Based Editing, click the filter icon, and choose the filler words option. Premiere highlights hesitations like um, ah, and uh in the transcript, and deleting them from the transcript removes them from the sequence. It needs a Creative Cloud subscription, and Adobe has been clear that detection is not perfect.

Does removing filler words make speech sound choppy? expand_more

Not if the cuts are padded. VidClean pulls each cut inward by 40 milliseconds so neighbouring words keep their edges, and fades each seam over 8 milliseconds so the join makes no click. Back-to-back fillers such as "um, uh" merge into a single cut instead of two.

Does it also remove my pauses? expand_more

No. Filler removal cuts only the filler sounds themselves, never silence. Intentional pauses stay where you left them. If you also want dead air tightened, run silence removal, which is a separate pass.

Can I remove filler words from audio only, like a podcast? expand_more

Yes. Upload an MP3, WAV, or M4A. There is also a dedicated filler word tool that returns a cleaned MP3 rather than a video. The flagged transcript there is free; the cleaned MP3 download is a Pro feature.

Does it work in languages other than English? expand_more

Yes. Word-level timestamps are available for 25 languages, including Spanish, Portuguese, French, German, Italian, Dutch, Polish, Ukrainian, and the Nordic languages. The hesitation sounds VidClean matches are the ones these languages share.

Related reading

How to remove silence from a video for free. The other half of tightening a raw recording.

Text-based video editing, free. Editing video by editing its transcript.

How much of a recording is actually silence. Measured across VidClean's own uploads.