How Automatic Silence Removal Works
What this tool does
Recordings of talks, podcasts, interviews and voice memos are full of dead air: the seconds before someone starts speaking, the pause while a guest thinks, the gap after a slide change. This tool finds those quiet stretches and joins the parts that remain, so a 40-minute lecture with long pauses can become a tighter 32-minute file without editing it by hand.
Everything happens inside your browser. Your audio is decoded and processed on your own device; nothing is sent to a server, and there is no account, watermark or limit on how many files you clean up.
How silence is detected
The audio is split into 20-millisecond frames and the loudness (RMS level) of each frame is measured. A frame counts as silent when it is quieter than the silence level you set — for stereo files the louder channel decides, so a sound on only one side is never cut. A run of silent frames becomes a cut only if it lasts at least as long as the shortest pause you chose.
Padding then hands back a little silence on each side of every cut, so the last syllable before a pause and the breath after it stay natural. Where two pieces are joined, an 8-millisecond crossfade smooths the edit so you don't hear a click.
Example: tightening a recorded lecture
A 45-minute classroom recording has a steady hum from the projector, so its quiet parts sit around −52 dB. With the default −40 dB level the pauses between explanations are found easily; the summary shows 210 silent parts and a result about 7 minutes shorter. Raising the shortest pause to 1,200 ms keeps the speaker's natural rhythm and only removes the long waits while they write on the board.
After pressing Remove Silence, the result plays right in the output panel. If it sounds rushed, add 50 ms of padding and run it again — the waveform updates as you move each slider, so you can see the effect before processing.
Notes and limits
The tool removes pauses, not filler words: an "um" or a cough is louder than silence and stays. Very noisy recordings may need a higher silence level, and music with quiet passages can lose intentional rests, so check the red parts before you export. Files up to 500MB are accepted, but long WAV files use a lot of memory; if your browser runs out, split the recording first with the audio trimmer. Tags such as title and cover art are not copied to the new file, and the sample rate follows your device's audio setting (usually 44.1 or 48 kHz).