Automatically shorten long pauses
The detector examines the loudest sample in each short window and groups nearby audible windows into kept sections. Gaps longer than the selected minimum are removed; natural brief pauses stay in place.
Choose a silence threshold in dB
A lower value is more sensitive. At −50 dB, very quiet room noise may count as audio. At −30 dB, soft speech may count as silence. The −40 dB default is a practical starting point for spoken recordings, but previewing before download is important.
Keep padding for natural edits
Padding retains a small amount of quiet audio before and after every kept region. Overlapping padded regions are combined, so audio is never duplicated. The waveform shows planned cuts before export; listen to the processed preview before downloading. This is threshold-based editing, not AI speech detection or noise removal. For manual edits, use the audio trimmer.
Frequently asked questions
How does the audio silence remover work?
It measures the signal in short windows, keeps audible regions, joins speech or music separated by short pauses, and cuts gaps longer than your minimum-silence setting.
What silence threshold should I use?
Start at −40 dB. Move toward −30 dB for noisier recordings or toward −50 dB when quiet speech is being removed.
Is the audio uploaded?
No. Detection, editing, preview, and WAV encoding all happen locally in your browser.