What is the padding for?

It keeps fifty milliseconds of audio either side of each cut so the two surviving ends do not meet at a level mismatch, which would be heard as a click. It is short enough that you do not hear it as a fade, but it is the reason the joins pass unnoticed.

Which threshold preset should I use?

The one that matches the room: the quiet setting for a treated room or a close microphone, the medium one for an ordinary living room, and the loud setting where there was air conditioning or traffic. Choosing a setting quieter than the actual room leaves gaps behind; choosing a louder one starts eating quiet speech.

Why does the minimum length matter so much?

Because it is what distinguishes a pause from a breath. Set too short, the natural gaps inside and between words are removed and the speech sounds rushed and clipped. It is the control to adjust first when too much has been taken out.

Does it make the recording quieter?

No, it makes it shorter. The level of everything that survives is unchanged and the silence is deleted rather than reduced, which is why the reported removal is a duration and not a level change.

Can I undo it if too much was removed?

Not from the result, which no longer contains the removed audio. The original file is untouched and is the thing to process again with a lower threshold or a longer minimum. Running the tool twice on the same file is not needed, since the whole file is processed in one pass.

Silence Remover

You Might Also Need

Audio Splitter

Split audio files into multiple segments

Silence Statistics Calculator

Calculate silence statistics

Audio Trimmer

Trim audio files to keep only the parts you need

Noise Reducer

Reduce background noise in audio

Silence Remover finds the quiet stretches in a recording and cuts them out, leaving the material that matters. Each cut keeps a little padding so the joins do not click.

What is Silence Remover?

Choose a threshold and a minimum length, drop a file, and process. The footer reports how many seconds of silence were removed across how many cuts.

The controls:

Three threshold presets, labelled by the room they suit: a quiet one at fifty decibels below full scale, a medium one at forty, and a loud room at thirty.
A minimum silence length from two hundred to two thousand milliseconds, in steps of a hundred, defaulting to seven hundred.
A footer line giving the removed time, the number of cuts, and the padding kept at every edge.
Detection runs on the first channel, but every channel is cut at identical positions so a stereo file stays aligned.

The two settings are doing different jobs and both matter. The threshold decides what counts as silence at all: a low setting only treats the very quietest material as a gap, so it suits a recording made in a quiet room, while a higher setting is needed when the room had air conditioning or traffic, at the cost of also treating quiet musical passages as gaps. The minimum length decides how long a quiet stretch has to last before it is considered removable, and it is the control that protects the natural pauses in speech. The detection works on short frames and measures the level within each one, keeping runs of quiet frames that last at least the minimum; because it is a sustained measurement rather than a per-sample decision, a single quiet sample in the middle of a word never becomes a cut. What makes the result listenable is the padding. Fifty milliseconds are kept at each end of every cut, so the material either side does not butt together at a level discontinuity, and the transition is short enough to be inaudible as a fade but long enough to avoid a click. The arithmetic of how much comes out is reported rather than guessed, which matters when the purpose is to shorten a recording to a fixed length: run it, read the figure, and adjust the minimum length rather than the threshold if too much or too little has gone.

What it does not do:

It cuts rather than compresses, so the timeline changes length.
Only the level is considered, so a quiet passage of wanted audio is treated as silence.
Detection uses one channel, so a gap in one channel only is not found.
The threshold presets are three fixed levels rather than a free value.
The removed audio is not recoverable from the result; only the original still holds it.
Output is sixteen bit WAV.

Use it to shorten a lecture that is full of pauses, to tighten a spoken take before transcription, or to strip the dead air out of a field recording with long gaps between events. When the aim is to divide a recording at its silent gaps rather than remove them, the splitter does that, and when the quiet stretches should be pushed down rather than taken out, the gate and the expander reduce them while keeping the timeline intact.

How to use Silence Remover

Drop an audio file onto the panel and note the duration.
Pick the threshold that matches the room the recording was made in.
Set the minimum silence length so that natural speech pauses are not cut.
Press process, read the removed time and cut count, and download the shortened WAV.

When to use Silence Remover vs related tools

Silence Remover is the tool for making a recording shorter by taking out its dead air, which is a different job from making it quieter. Audio Splitter uses those same quiet gaps as cut points and keeps every section as its own file instead. When the pauses should stay in place but sit lower, Audio Gate and Audio Expander reduce them, and Audio Trimmer removes a single span you name yourself. To measure the gaps before cutting them, Silence Statistics Calculator reports how much silence a file holds and where it sits.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.