Remove silence from a video automatically

How the silence detection works

Vidmini measures the loudness of the sound in small windows of 20 milliseconds, from the first second to the last. Every stretch that stays below the chosen “Silence level” for longer than the pause length you picked is marked as silence. The video and the audio of those stretches are then removed together and the remaining parts are joined, so lips and voice stay in sync.

The detection is based on loudness only. It is not AI and it does not understand words: it won’t remove filler sounds like “um” or “uh”, and a quiet room is what makes the cuts possible. Background music or constant noise, such as a fan or traffic, can keep the level above the threshold and prevent cuts.

The three settings

“Silence level” goes from −60 dB to −20 dB, with −40 dB by default. Toward “Only true silence”, only near-total silence is cut. Toward “Also quiet sounds”, breathing, room tone and soft noises are cut as well. If your recording has a noticeable background hum, raise the level.

“Remove pauses longer than” sets the shortest pause that is cut: 0.3, 0.5 (default), 0.8 or 1.5 seconds. Short values give a fast, dense rhythm; long values only remove the real gaps and keep natural breathing pauses.

“Keep around speech” leaves a margin of 0.05, 0.15 (default) or 0.3 seconds before and after each sound, so the first and last syllables are not clipped. A larger margin sounds more natural, a smaller one gives snappier jump cuts.

For talking-head videos, tutorials and podcasts

Cutting pauses by hand in an editor means scrubbing through the timeline and splitting clips one by one. For a talking-head video, a screen tutorial, a recorded lecture or a vlog, this tool does that first rough cut in one pass and gives you the classic jump-cut style used on YouTube, TikTok and Reels.

It works on audio files too. Drop a podcast episode, an interview or a voice memo and you get an M4A file without the dead air. For videos, the output keeps the container of the source: an MP4 stays an MP4, a MOV stays a MOV. Picture and sound are re-encoded in a single pass, so the result is not lossless, but on a modern device the difference is rarely visible.

Private and free

Your recording is analyzed and cut by your browser, using the WebCodecs API and WebAssembly. Nothing is uploaded, there is no account and no watermark, and once the page has loaded it works offline. There is no size limit set by us; on phones, free memory and storage decide how long a recording you can process.