What does a silence remover do?
A silence remover scans a recording for stretches where the sound stays below a set level for longer than a set time, then cuts those stretches out and joins the rest. On a talking-head video the result is a jump-cut edit: the same speech, delivered without the pauses, hesitations and dead air between sentences.
It works purely on loudness, so in practice it acts as a simple jump cut editor that is fast and predictable: anything quiet and long enough is removed, anything spoken stays. It also means the settings matter more than any clever algorithm, which is why this page shows the cuts on a waveform before anything is exported.
How the pauses are found and cut
When you choose a file, the tool makes a small mono copy of the sound at 8 kHz, which is plenty to judge loudness and quick to read even for an hour-long recording. It splits that copy into 20 millisecond windows and measures the level of each one in decibels.
To pick a starting threshold, it sorts those levels and takes the quietest tenth of the recording as the room's noise floor, then sets Silence below 8 dB above it. Every run of windows under that line that lasts longer than Remove pauses longer than becomes a cut, shortened at each end by half of Leave this much pause. Silence at the very start and end is removed completely.
On export, FFmpeg keeps only the parts to keep and re-encodes the video as H.264 with AAC sound, a combination that plays on every phone, browser and editor. Audio-only exports are 16-bit WAV or 192 kbps MP3. Audio files such as MP3, WAV and M4A work the same way and export as audio.
Jump cuts: the look of a video without pauses
Removing pauses from video produces jump cuts, where the speaker's head shifts slightly between one sentence and the next. On YouTube, TikTok and Reels that look reads as energetic and deliberate, and it keeps viewers listening. A few habits make it look intentional rather than accidental:
- Keep a little pause. 0.1 to 0.2 s at each cut sounds like natural speech. At 0 s, sentences run into each other.
- Use a locked-off camera. A tripod keeps the frame identical, so only you move at each cut.
- Hide big jumps. In your editor, cover a jarring cut with B-roll, or punch in to a tighter crop on alternate sections.
- Raise the minimum pause for teaching videos. 1 s or more keeps the thinking time that helps a lesson land.
Removing filler words like um and uh
A filler word is speech, so its level sits above the silence line along with every other word. Filler words are found by reading the transcript, and the free way to do that is a short workflow: run the silence remover first, transcribe the shorter video with the captions tool in your editor or any transcription service, then read through the text. The SRT to TXT converter turns a subtitle file into a plain transcript that is easy to scan, and every timecode in the original SRT points you to the exact spot.
From there, cut the ums in CapCut or in Premiere Pro's text-based editing, where deleting a word in the transcript removes it from the timeline, or give the transcript and the video to an AI assistant to produce a cut list. If you keep the captions, the SRT editor shifts the timing to match the final edit.
Where removing dead air helps most
- Talking-head videos. Tutorials, product demos and vlogs recorded in one long take.
- Podcasts and voice-overs. Export WAV for further editing, or MP3 to publish straight away.
- Lectures and webinars. Long gaps while slides change or questions come in.
- Screen recordings. The pauses while a page loads or you look for the next click.
The silence remover works in current browsers on desktop and on phones. A laptop or desktop handles long, high-resolution videos best, because the re-encode runs on your own processor. When you also want the untouched full soundtrack as a backup, the audio extractor saves it as MP3, WAV or M4A.