How to remove ums and filler words from a video
By Anteras · Updated
To remove ums from a video automatically, the video is transcribed with the timing of every word, the filler sounds are found in that transcript, and the picture and sound are cut at exactly those moments. Cut the sounds (um, uh, erm) and leave filler words (like, you know, so) alone unless you check each one, because those are often part of the sentence.
Filler sounds and filler words are not the same
| Kind | Examples | Cut automatically? |
|---|---|---|
| Filler sounds | um, uh, erm | Yes. They carry no meaning and the sentence is whole without them |
| Filler words | like, you know, so, basically | Not without checking. “I like this one” needs its “like” |
| Repeats and restarts | “I think, I think we should” | By hand. Software cannot know which attempt you meant |
Anteras removes the first kind only. That is a limit, and it is on purpose: an automatic cut that takes a real word out of a sentence is worse than an um left in.
How automatic removal works
- The speech is transcribed, with a start and end time for every word.
- Each word is compared with a list of filler sounds for the language spoken.
- The video is cut at the timings of the matches, and the picture and sound on either side are joined.
- Captions are placed afterwards, so they stay in time with the shortened video.
The list depends on the language
| Language | Sounds cut |
|---|---|
| English | um, uh, erm, uhm |
| German | äh, ähm, öh, öhm, hm |
| Spanish | eh, em, ehm, mmm |
| French | euh, heu, hum |
A separate list per language matters more than it seems. “Um” is a filler in English and an ordinary word in German; one shared list would cut real words.
When a real word is heard as a filler
It happens: a short word, mumbled, is transcribed as “um”. In Anteras the fillers that will be cut are struck through in the Words list before you export. Tap one, type what you actually said, and it stays in the video and in the captions.
Doing it in Anteras
- Upload your video.
- Switch on Remove filler words.
- Open the Words list and look at what is struck through.
- Correct any real word that was heard as a filler.
- Watch the free preview, then export in 1080p.
The tool page: remove filler words.
Doing it by hand
In a timeline editor, play the video, stop at each um, and cut on both sides of it. Zoom into the waveform so the cut lands in the gap before the next word and not on its first consonant. It works, and it is slow, which is why most people stop after the first minute.
Pauses are a separate job
Cutting an um often leaves the pause that surrounded it, and the video still drags. Pauses are found differently, from the sound level and not from the transcript, so they are a second switch: see how to remove silences from a video automatically. Use both and the take tightens up without sounding chopped. For the whole order of an edit, see how to edit a talking-head video.
Saying fewer of them
The cheapest um to remove is the one you never said. Two habits help, and neither needs talent: know your first sentence before you press record, and when you lose your thread, stop and say nothing. A silent pause is cut cleanly; an um in the middle of a word is not.
Questions
How do I remove ums from a video automatically?
Use a tool that transcribes the video with word timings and cuts the filler sounds at those timings. In Anteras that is the Remove filler words switch.
Does it remove “like” and “you know”?
Not in Anteras. Only filler sounds are cut, because those words are often part of the sentence.
Will the cuts be visible?
Each cut is a small jump in the picture, as with any cut in a single take. Viewers of short video are used to them; keeping still while you think makes them smaller.
Is it free?
Yes. Filler removal is on the free plan: 3 exports a month, videos up to 3 minutes, 1080p, no watermark.