Cut Silence from Video & Audio
Finds the silent or voiceless stretches in a video or audio file and cuts them out. Long pauses in commentary and lecture recordings get trimmed away in no time, right in your browser.
Your files are never uploaded. Everything runs on your device.
How to use
- 1Pick a fileDrop or click to select a video (MP4 / MOV / MKV / WebM) or audio file (AAC / M4A / MP3 / WAV / FLAC)
- 2Check what it foundLook at the waveform, the cut ranges, and how much shorter it gets. Switch modes if it does not look right
- 3Run and saveCheck the cut preview, then save the file
How this tool works
The tool first turns the audio of your video or audio file back into a waveform. It then looks for stretches with no sound or no voice, removes them, joins what is left, and packs it into a new file. It is for tightening the thinking pauses in commentary and lecture recordings, or the dead air in the middle of a voice recording.
There are two ways to find those stretches, and they look at different things.
| Silence mode | Voice mode | |
|---|---|---|
| What it looks at | Volume | The volume of the voice, after an AI that picks out speech has suppressed music and background noise |
| How it decides | It sets a threshold from the quiet parts of the whole file and takes any stretch that stays below it for long enough | The same rule, applied to the voice-only signal |
| Best for | Recordings where the sound stops when nobody is talking | Recordings with music or background noise playing throughout |
In both modes, a little margin is left around each stretch so that the start and end of words are not clipped. AAC and Opus audio is cut at the boundaries of the compressed data, so it is not re-encoded.
The AI used by voice mode is a small model that has already finished learning. It never learns from your voice. It is downloaded from this site once, the first time you use it, and runs on your device. That download is the only network traffic: neither your file nor the result of the detection is sent anywhere.
Things to know
- If the original video has subtitle tracks, they are not carried into the output and are lost.
- If the original video has transparency (an alpha channel), the transparent areas in the output are filled in, usually with black.
- The preview and the output can differ slightly, by an amount you will not notice.
- If a video has several video tracks, only the first one is output. If it has several audio tracks, the stretches are found on the first one and all tracks are cut at the same positions.
- For audio-only files, the output format is the same as the input.
- Voice mode cuts anything without a voice, including scenes with only game audio or sound effects. Loud background music lowers its accuracy.
- Surround audio with three or more channels cannot be processed when conversion is needed.
Last updated: