Skip to content

Remove Silence from Audio Online — Free Trimmer

Cut the dead air at the ends of a recording, or every pause inside it, and see how many seconds went.

ENGINEFFMPEG SILENCEREMOVE
ACCEPTSMP3 WAV OGG M4A FLAC AAC OPUS WMA
MAX SIZEMEMORY-BOUND
UPLOADNEVER

Nothing is enforced, but the waveform is decoded to 32-bit float in this tab at roughly 20 MB per minute of 44.1 kHz stereo, and the trimmed file is built whole in the engine's memory before it is handed back.

01Upload your audio file by dragging it into the drop zone or clicking to browse.
02Choose Ends only or Every gap under WHAT TO CUT, set how quiet counts with the Silence below slider, then click Remove silence.
03Check how many seconds went, then click Download to save the trimmed audio.

About Remove Silence

Recordings arrive with dead air on them. There is the pause before you started talking and the one after you stopped, and then there are the gaps in the middle — the moment before an answer, the space where somebody was thinking. This tool cuts either. Ends only trims the quiet run at the start and the quiet run at the end and leaves everything between them exactly as it was, pacing included; it is the mode for tidying a recording without editing it. Every gap removes each quiet stretch longer than the minimum gap wherever it falls, which shortens the file and changes its rhythm — pauses between sentences close up. That is what you want for a rough cut and not what you want for music. Two settings control what counts as quiet. The threshold is how quiet a stretch has to be: −50 dB catches room tone, and moving it nearer zero starts catching quiet speech and breath along with it, which also risks cutting into words. The minimum gap, which only matters in the second mode, leaves anything shorter alone — below about a third of a second you are closing the spaces between words rather than between sentences. The cutting is done by FFmpeg's silenceremove filter running in this tab, and the three durations on screen are FFmpeg's own report of what it read and what it wrote, so they carry its hundredth-of-a-second resolution rather than sample precision. The output sample rate is the source's on the WAV and FLAC arms — nothing here asks for a resample. MP3 is the exception it cannot help: MPEG layer III tops out at 48 kHz, so the encoder writes a 96 kHz source at 48 kHz, and the run reports it when that happens. Tags and cover art are not carried across: the picture stream is not mapped and the metadata is explicitly dropped rather than copied to the new file.

Questions

What is the difference between the two modes?

Ends only trims the quiet run at the start and the quiet run at the end and leaves everything between them exactly as it was, pauses included — it tidies a recording without changing its pacing. Every gap removes each quiet stretch longer than the minimum gap wherever it falls, which shortens the file and closes the pauses between sentences.

What should the threshold be?

−50 dB is the default and catches room tone. Moving it nearer zero starts catching quiet speech and breath as well, which also risks cutting into words; moving it further down catches only near-digital silence. If the tool removed something you wanted, lower the threshold before changing anything else.

Will it cut in the middle of a word?

It can, if the threshold is set high enough that quiet speech reads as silence. That is what the minimum gap is for in Every gap mode: a quiet stretch shorter than it is left alone, and below about a third of a second you are closing the spaces between words rather than between sentences.

How exact are the durations it reports?

They are FFmpeg's own report of what it read and what it wrote, which carries hundredth-of-a-second resolution — treat them as accurate to about a twentieth of a second rather than to the millisecond. Removed is the difference between the two, and it never goes negative: a file whose header duration is missing reads as zero rather than as a nonsense number.

Does it change the sample rate or keep my tags?

On WAV and FLAC the sample rate is the source's — silenceremove was measured leaving 44.1 kHz alone, and no resampling is asked for. MP3 cannot go above 48 kHz, so a 96 kHz source is written at 48 kHz by the encoder itself; the page reports the two rates when they differ. Tags and cover art are not carried over, because the metadata is explicitly dropped rather than copied to the new file.

Is my file uploaded to a server?

No. Transmute processes everything locally in your browser using JavaScript and WebAssembly. Your files never leave your device — there is no server, no upload, no cloud processing.

Related