How do I remove filler words from a video?

Upload the video to Valmera and type "remove the ums and uhs". The AI agent cuts every um, uh, er and hmm at its exact word timestamps in one pass, renders a preview, and reports what it removed. Review it, ask for changes in chat, then export a full-quality MP4 rendered from your original file.

Which filler words does Valmera remove by default?

The default set is the non-word hesitations: um, uh, er and hmm. They are safe to cut everywhere because they never carry meaning. Name any other word or phrase, such as "you know" or "basically", and the agent cuts those too.

Can I remove "like" and "you know" from a video?

Yes. "You know" and "basically" can usually go everywhere. "Like" and "so" are real words as often as fillers, so ask for the filler uses only: the agent reviews each instance in the transcript and cuts the ones that are hesitations, leaving "I like it" alone.

Will cutting filler words make the video sound choppy?

Each cut lands on the filler's own word boundaries, so the words around it stay whole. Speech can feel rushed if a speaker leans on fillers heavily; keep a few ("leave the hmm at 3:05") or restore any cut after you hear the preview.

Will my captions show the ums?

No. Valmera's captions skip filler sounds, so removing them cleans up the audio and the pacing. If you add captions later, they are timed to the edited video.

Can I remove filler words and silences at the same time?

Yes. "Remove the filler words and the dead air" runs both cleanups and returns one preview. Removing a filler leaves the pause around it, so pairing the two is what makes speech sound tight.

Is the filler word remover free?

Creating an account and uploading are free. Editing requires a subscription. Exports are full-resolution H.264 MP4s and end with a brief Valmera end card.

Remove Filler Words from Video

To remove filler words like um and uh from a video, upload it to Valmera and type “remove the ums and uhs.” The AI agent cuts every um, uh, er and hmm at its exact word timestamps in one pass, and you can name any other crutch words, like “you know” or “basically,” to cut along with them.

The cuts are non-destructive. You review a preview, put back any filler you wanted to keep, and export a full-quality MP4.

Cut the Ums in One Request

Upload, type one sentence, review the preview. Creating an account and uploading are free; editing requires a subscription.

Create an account →

See pricing →

How to Remove Filler Words in Valmera

1
Upload the video
MP4, MOV, M4V, MKV or WebM, up to 14 GB and 3 hours. Indexing builds a word-level transcript that keeps every um and uh with its own timestamps.
2
Ask for the cleanup
Type “remove the ums and uhs.” Add your own crutch words (“and every ‘you know’”) or anything to keep.
3
Review the preview
The agent cuts each filler on its word boundaries, checks what the edit still contains, and renders a preview. Listen to a few joins.
4
Adjust and export
Ask for any filler back (“keep the hmm at 3:05”) or tighten the pauses too, then export a full-resolution MP4 from your original file.

PROMPTS TO PASTE

Remove all the ums and uhs.

Cut the ums, and every “you know” too.

Cut “like” only where it’s used as a filler.

Remove the filler words and tighten the pauses, but keep my “hmm” before the answer at 3:05.

Why the Transcript Has to Keep the Ums

Many speech-recognition systems tidy filler words out of the transcript, the way a human captioner would. That reads well and makes filler removal impossible: you cannot cut an “um” the transcript never wrote down. Valmera's transcription keeps every hesitation, each with its own start and end time, so the cut removes exactly that sound and nothing either side of it.

That is also why this beats cutting by volume. Fillers are short, quiet and glued to the words around them, which is the worst case for a waveform threshold. The word-level timestamps put the edge on the boundary between words instead.

Settings That Matter

Setting	Default	How to change it
Filler set	um, uh, er, hmm	“Also cut every ‘you know’ and ‘basically’.”
Real words used as fillers	Not cut	“Cut ‘like’ only where it’s a filler.” The agent reviews each use in the transcript.
Keep a few	None kept	“Keep the hmm before my answer at 3:05.”
Scope	Whole video	“Only clean up the intro.”
The pause around each filler	Left as is	Add “and tighten the pauses” to remove the gap too.

Breaths are not fillers. The breath before a strong line is part of the delivery, so the filler pass leaves breaths alone and the silence pass keeps a little air around speech.

Common Mistakes

Treating every “like” as a filler. Ask for the filler uses only, or the meaningful ones go too.
Stripping hesitation that carries meaning. In an interview, a “hmm” before an honest answer reads as thought. For a natural or raw feel, ask for a lighter pass.
Cutting fillers but not the pauses. A removed “um” often leaves a gap. Ask for both: “remove the fillers and tighten the pauses.”
Judging from the summary alone. Listen to a few joins in the preview, especially fast speech.

Limits

The pass removes what the transcript recognized. A mumbled or overlapping “um” can be missed; give the agent the timestamp and it cuts that moment directly. Cuts are hard splices with no crossfade, and English is the best-tested transcription language. There is no audio denoise.

Part of a Full Cleanup

A clean talking-head edit is usually three passes: bad takes, dead air and fillers. One request runs all three, and captions can follow in the same chat. Podcasters can use the combined podcast cleanup. More workflows are on the podcasters, YouTubers and course creators pages and in the cuts and cleanup docs; the AI video editor page covers everything else.

Using Claude, ChatGPT or Cursor? The same pass is the remove_filler_words tool on Valmera's MCP server.