> For the complete documentation index, see [llms.txt](https://learn.firecut.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://learn.firecut.ai/features/remove-filler-words.md).

# Remove filler words

## ❓ What is it?

Remove filler words helps you find words like "umm" and "uhh" in your sequence that are typically unwanted, and then helps you cut them out.

## 🤔 How does it work?

This feature:

* Finds the filler words in your sequence when you click **Analyze audio**, by listening to the guidance audio tracks you pick
* Shows these filler words in context of the rest of your sequence. You can right-click words to select/deselect them for deletion
* Lets you place markers in the sequence where these words occur with **Add markers to timeline**, if you only want to mark them and double check them later
* Cuts them out of ALL tracks when you click **Cut selected words**

{% embed url="<https://www.youtube.com/watch?v=J9s4H-p7bs8>" %}

## ⚙️ Settings

There are 3 main settings for remove filler words:

1. **Filler words guidance audio track(s):** The audio tracks you want FireCut to listen to for filler words. Only select the tracks with your speaker's voice. The filler words you choose are cut from ALL tracks, not only these.
2. **Scope:** You can decide whether to run the operation on your **Full sequence** or only on a certain portion that you specify with **In / Out points** (you can place these in your sequence using the `I` and `O` hotkeys). This starts on **In / Out points**, and your scope needs at least 5 seconds of audio.
3. **Language:** The language spoken in your audio: English, Dutch, French, German, Italian, Polish, Portuguese, Spanish or Turkish. FireCut also uses it to pick the filler words it starts with. We are constantly adding support for new languages. However, please note this is experimental and may create unpredictable results when compared with English. This is the case for every AI tool even if they don't mention it, because [training sets for AI tend to be in English](https://www.wired.com/story/chatgpt-non-english-languages-ai-revolution/).

## ⚒️ Reviewing the results

After the analysis, FireCut shows your transcript with the filler words marked for removal. Before you cut, you can adjust:

* **Timing adjustment:** **Padding (start)** and **Padding (end)** go from -500 to 500 ms and start at 0 ms. Increase them to cut less and make the cut tighter. Reduce them to cut a bit extra around each filler word. They're linked, so they move together: click the link icon to set them separately. Select a word in the transcript first, then change the padding to see how it affects the cut.
* **Bridge empty space between words:** On by default. FireCut stretches each word to the middle of the pauses on either side, so cutting a filler word also takes the pause around it, and back-to-back filler words are cut in one go.
* **Filler words:** The words FireCut looks for. It starts with the usual ones for your language (in English, "um" and "uh"). Click **+** to add your own word or phrase, and hover over a word and click **x** to remove it. Words you add count for this analysis only, and are cleared when you click **Analyze audio** again. Words FireCut picked out from the context of your sentences, like a "you know" used as a filler, have an outline. Removing one of those unmarks the words it found.
* **Results:** Left-click a word to go to it in your timeline. Right-click a word to mark or unmark it for removal.

Then:

* **Add markers to timeline:** Adds a "Filler word" marker at the middle of each marked word, so you can check them in your timeline later.
* **Cut selected words:** Cuts the marked words out of ALL tracks.

## 💡 Tips for best outcomes

* **Cut out any silences first**, and make sure any music / sound effects are muted
* **Remove any repetitions first**, as FireCut might not transcribe multiple instances of the same text
* **Don't work in the timeline** while FireCut is working in the background
* **Don't move clips between the analysis and the cut.** FireCut's transcription follows your **sequence**'s timecode, not individual clips, so moved clips put the cuts in the wrong place

## 🙋‍♂️ Common questions, issues, watch-outs

* **"It did not find all the filler words"** ->
  * This feature is currently in beta and might miss some instances of filler words. If you do find that the detection is not accurate, please report it to us so we can fix it :) -> <support@firecut.ai>
