Video, Audio & YouTube
Remove silence and long pauses from audio
Upload a WAV, MP3, M4A, FLAC, MP4 or MOV recording. You first see every detected pause on a waveform, and nothing is cut until you confirm the list.
or drop it here
- WAV / MP3 / M4A / FLAC / MP4 / MOV
- Up to 30 minutes
Detect two candidate ranges, review them, then explicitly confirm those exact ranges in a second step
- The first step creates a cut-list and waveform without editing media
- The render confirmation is bound to the exact source hash, bytes, policy and ranges
- Applied cuts remain preserved as JSON and CSV
detected-cuts.json · detected-cuts.csv · source-waveform.png · manifest.json
One clear job, from source to download
- 1
Add the source
Supported formats and limits are visible before the upload.
- 2
Confirm the settings
Review the exact source, options, units and access before processing.
- 3
Inspect and download
Check the preview and warnings, then unlock the complete package.
Finding and cutting long pauses in a recording
What you upload and how pauses are found
You upload one recording in WAV, MP3, M4A, FLAC, MP4 or MOV, up to 30 minutes long. The tool measures the audio level and marks ranges that stay quiet long enough to count as a pause. It does not listen for words, so fillers such as um and uh are treated as speech and stay in. A quiet passage is only reported when it stays below your threshold for at least the minimum pause length.
Two steps: review the cuts, then render
The first step does not touch your media. You get the detected cuts as JSON and CSV plus a waveform image of the source, so you can see each range before anything is removed. When you confirm that exact list, the second step renders the edited file with a waveform, an edit decision list and a manifest. If the source file changes in between, the render is refused instead of cutting at the wrong timestamps.
Settings and your judgement
Three settings control the result. Silence threshold sets how quiet a range has to be. Minimum pause sets how long it has to last. Keep padding leaves a set number of milliseconds of the original quiet on each side of a cut, so speech does not start abruptly at the join. A measured silence is not proof that a pause is unwanted. Some pauses belong in the recording, so you decide which ranges go.
Questions before you run it
Why does the first step not edit my audio?
The detection pass only measures levels and writes a cut list plus a waveform image of the source. Nothing is re-encoded until you confirm the exact ranges in a second step.
How is a silent range decided?
By measured level against the silence threshold you set, combined with the minimum pause length. A quiet passage is only reported when it stays under the threshold for at least that duration.
Can it cut filler words such as um and uh?
No. Detection is acoustic, so a spoken filler is audible content and will not be flagged. Only measured quiet ranges reach the cut list.
What does keep padding do?
It leaves a set number of milliseconds of the original quiet at each side of a cut, so speech does not start abruptly at the join.
What if the file changes between detection and render?
The render confirmation is bound to the source hash, byte count and the exact ranges from the preview. A changed file no longer matches and the render is refused instead of cutting at the wrong timestamps.