Only Trusted Silence Remover
Cut the dead air out of a recording, and see every cut before it happens. About a quarter of a typical recording is silence; this shows you yours. Your audio is never uploaded.
Drop an audio file here, or tap to choose
Podcasts, lectures, voice memos, interviews. MP3, WAV, M4A, OGG, Opus, FLAC and WebM all decode in your browser.
Start from a preset
Presets set the three sliders. The sliders stay in charge after.
Pauses are shortened to the "leave" length, never removed entirely, so word endings and breaths survive. Changes redraw the cuts instantly.
This file
Decoding...
Orange bands are the pauses that will be shortened. Click a band (or use the list below) to give that pause back; click again to cut it again. The cleaned preview plays the exact joins you will export, micro-fades included.
Export the cleaned audio
Check the result
DownloadProcessed on this device: 0 · Uploaded: 0
The cuts
How it works
The cuts are drawn before anything is removed
Your browser decodes the recording once and measures the loudness of every 10 milliseconds. A run of quiet at least as long as your "shortest pause" setting is a candidate cut; the middle of the run is removed and your "pause to leave" stays in place, split between the end of one phrase and the start of the next. At every join the export applies an 8 millisecond fade, so joins do not click. Moving a slider re-plans the cuts from the loudness measurements in a few milliseconds, which is why the waveform bands and the scoreboard update instantly, and why you can play the cleaned version and hear exactly what will be exported before you commit to it. Tapping a cut gives that single pause back at full length.
Guide
Getting a clean result
- Start with Podcast, then listen. Play the cleaned version and listen to a few joins. If a word ending feels clipped, lower "how quiet counts as silence" by 3 dB or add a little more "pause to leave".
- One bad join is not a settings problem. If a single pause was judged wrong, give just that one back from the cut list instead of loosening every setting.
- Noisy rooms: fans and traffic raise the quiet floor, so silence stops looking silent. Raise the threshold a few dB (toward -32) until the bands land on the real pauses.
- Music and performances: pauses are part of the piece. Use Gentle, or skip this tool for anything where timing is the art.
- Long lectures: export WAV only if an editor needs it; MP3 at 192 kbps is about 1.4 MB per minute and nothing in a lecture needs more.
Questions
Will it cut the ends off my words?
That is the known failure of silence removers, and this tool is built against it. Pauses are shortened to the length you choose, never removed entirely, so speech keeps breathing room on both sides. Every cut is drawn on the waveform before anything is exported, you can play the cleaned version and hear the actual joins, and tapping any cut gives that one pause back without touching the others.
What settings should I use for a podcast or lecture?
Start with the Podcast preset: it treats anything under -40 dB as silence, only touches pauses of 0.6 seconds or longer, and leaves 0.35 seconds of every pause in place. For a quiet room and a careful speaker, Gentle (-45 dB, 1 second, leave 0.5) is safer. For fast jump-cut style, Tight (-32 dB, 0.35 seconds, leave 0.2) removes more, but listen to a preview before exporting.
Should I export MP3 or WAV?
MP3 for sharing and publishing: much smaller, and 192 kbps is the default here. WAV when the audio goes into an editor next, because nothing is lost in compression. MP3 export is capped at 60 minutes of cleaned audio, the same cap the Audio Merger uses; WAV has no cap, but the file can be large (about 10 MB per minute of stereo).
Is my audio uploaded anywhere?
No. Decoding, analysis and encoding all run in your browser, on your device. Load the page, turn off your internet, and the tool still works; there is no server receiving your recordings.
Can it handle very long recordings?
The whole recording is decoded into memory, so very long or very large files can strain a phone. Files over about 350 MB get a warning before decoding. Around two hours of stereo is the practical ceiling on a desktop; on a phone, split long recordings with the Audio Trimmer first, or keep them under an hour.