Mute filler words EN

Found from the word-level transcript, which is why it needs captions first.

Rekorder can find the "um"s and "uh"s in a recording and mute each one. The picture keeps playing and only the sound goes quiet. It finds them in the recording's transcript, so the recording needs captions first.

Before you start

Generate captions for the recording. Rekorder looks for filler words in the transcript's word-by-word timing, so without captions it has nothing to search. See Generate captions for a recording.

Mute the filler words

Open the recording's video page and go to the Edit tab. Choose Remove silences and filler words. This opens the editor on the Audio tab. If you're already in the editor, open the Audio tab yourself. See What you can do in the editor.
Under Filler words, choose Mute filler words.
Wait a moment while Rekorder shows "Finding filler words…". When it's done, it tells you how many it muted, for example "Muted 12 filler word(s)." If there are none, it says "No filler words found."

Each filler word now has a muted span on the track. Press play to check them. Choose Apply & render when you're happy, because viewers only hear the change after you render. See Save a draft vs Apply & render.

You can run Mute filler words more than once. Rekorder skips any filler word that already has a mute, so you won't get duplicates.

What counts as a filler word

Rekorder mutes spoken hesitations, the sounds people make while they think. It recognizes these in English, German, French and Romanian, for example:

English: "um", "uh", "erm"
German: "äh", "ähm"
French: "euh"
Romanian: "ăă", "îî"

Longer versions such as "ummm" count too. Rekorder checks every recording against all of these languages, so an "um" in a Romanian recording is caught as well.

It doesn't mute real words that people use as fillers, such as "like" or "you know". It also skips agreement sounds like "mhm". To silence those, mute the span yourself. See Silence a section without cutting the picture.

Check and adjust the result

Remove a mute. Double-click the muted span on the track, or right-click it and choose Remove mute. Filler-word mutes work like any other mute.
Undo everything. Choose Undo in the Edit tab to take back the last action. One click of Mute filler words is one undo step.
Timing can be approximate. Hesitation sounds are hard to time exactly. When Rekorder can't tell where one starts and ends, it works out the span from the words on either side and never mutes more than 2.5 seconds for a single filler word. Listen to the spans if the speech around a filler is close together.
Silence detection is cleared. Adding mutes clears any silences you highlighted with Detect. Run Detect again if you still need them. See Remove all the dead air automatically.

Your original recording isn't changed. See Editing is non-destructive: getting the original back.

Troubleshooting

"Filler removal needs a word-level transcript"

The recording doesn't have a transcript with word timing. Generate captions first, then choose Mute filler words again. If the recording already has captions, generate them again. Some older transcripts don't include word timing.

"No filler words found"

Either there are none in the recording, or the transcript doesn't contain them. Captions sometimes leave out a hesitation the speaker did say. If you can hear one that wasn't found, mute it by hand.

The captions disappeared after I rendered

Rendering clears the captions, summary and chapters, because they no longer match the edited video. Mute the filler words before you render. See Why rendering clears captions, summary and chapters.