Choose a file to automatically detect and remove pauses, then preview the result.
Drop a recording here, or choose an audio file
Supported formats: MP3 · WAV · M4A · AAC · FLAC · OGGAll audio processing stays in your browser.
Zolkit Remove Silence is an online audio tool that detects sustained quiet intervals and joins the remaining audio into a shorter recording. Import a file and it automatically prepares a result to listen to. If the pacing needs work, adjust the detection settings and see the proposed cuts on the original waveform before applying them.
Choose an audio file or drop it into the tool. Detection and removal run automatically, and the Result tab opens when the audio is ready. Playback starts when you press Play.
Listen to the processed result, then switch to Original if you want to compare. Check the beginnings and ends of words, breaths and pauses between ideas.
Change a setting on the right. The tool switches to Original and marks the intervals that would be removed. The pause list and estimated output length update with the preview. Uncheck a pause in the list to keep it.
Click Apply settings & update below the sliders. It detects and processes in one step, then opens the new result. Listen again and use Export at the top right to download it.
The defaults are a starting point, not a rule for every voice. Change one setting at a time so you can see which cuts it affects.
Defines how quiet a frame must be to count as silence. Values closer to zero are more aggressive.
If pauses are missed, try raising it from −40 to −35 dBFS. If quiet speech is marked, lower it instead.
A quiet interval must last this long before it is eligible for removal. It does not change playback speed.
Increase it to leave short gaps alone. Decrease it to include shorter pauses.
Retains this much audio next to speech on each side of an internal cut, helping preserve word endings and breathing room.
Increase it if the result feels clipped or rushed. Decrease it for tighter edits. Too much padding can leave no interval to remove.
After a settings change, you are listening to and looking at the original audio. Shaded regions show the current removal plan; they have not yet been removed from a new result. Export becomes available after you apply the settings.
Click the ruler or waveform to position the playhead. Zoom in around a word ending and listen. Separate channel waveforms show sample peaks and RMS energy; waveform gain changes only the display, not detection or audio volume.
Open Review detected pauses to keep a deliberate pause by unchecking it. Those choices are included when you apply the current settings. Changing a detection parameter starts a new preview and resets earlier manual choices, so review them again.
The threshold may be too low for the room noise, or the minimum pause too long. Raise the threshold gradually or shorten the minimum, watching for quiet words that become marked. No detected pauses means the result stays unchanged.
Lower the threshold, increase the minimum pause or keep more audio at speech edges. You can also uncheck an individual interval. If the whole recording is marked, the tool blocks an empty result; keep an interval or use a lower threshold.
Keep more audio at speech edges or preserve selected pauses in the list. A shorter duration is not always a better edit: listen for pauses that separate thoughts.
Silence removal shortens quiet gaps. It does not clean noise underneath speech, separate speakers or remove background music. A noisy or music-backed recording may have no sufficiently quiet gaps.
Play a few edited transitions, including a quiet sentence and the start and end of the recording. Preview and export use the same processed audio. Short 3 ms fades at internal join edges help reduce clicks; they do not make an over-aggressive cut safe.
Download before refreshing, clearing or closing the page. Your original file is not overwritten; the download is a finished audio file, not a saved editing project.
This version uses RMS sound levels in 20 ms frames, expressed in dBFS. For stereo it uses the louder channel, so speech on only one channel is not treated as silence. It does not use VAD, recognize words or understand sentence boundaries.
Audio decoding, pause detection, preview, processing and encoding happen in the current browser. Audio content does not need to be uploaded to a server.
Common imports include MP3, WAV, M4A, AAC, FLAC and OGG when supported by your browser’s decoder. If a file cannot be read, try a compatible WAV file or another browser.
The layout supports phones, but decoded audio and the processed result use device memory. Capacity depends on the device and browser; there is no promise of unlimited file size. Use a computer, shorter recording or fewer open tabs if processing fails.
It removes selected time intervals and joins what remains, rather than speeding up speech or changing pitch. Processing and export re-encode the result; compressed formats can add lossy compression. WAV or FLAC avoid that additional lossy step.
Media, settings and results are held in the current page session. They are not saved as a reopenable project. Export the result you need before refreshing or closing the page.