Free tool · silenceremove · Runs in your browser
Remove silence from audio
Short answer
The tool removes every stretch of sound quieter than -40 dB, at the start, the end and in between. On 30 seconds of speech it removed 9.4 seconds (31.3%), leaving 20.6. But when a fan was audible in the pauses at only 15 dB below the voice, it removed 1.9%: noise above -40 dB is not silence. Nothing is uploaded.
Maximum file size: 100 MB. The first run loads a ~30 MB engine, once per visit. Nothing is uploaded - processing runs on your device.
Key takeaways
- On a clean speech recording the tool removed 9.389 of 30 seconds, 31.3%, so about a third of this narrator's recording was pauses below -40 dB.
- Background noise stops it. With a fan added 15 dB below the voice (-38.1 dBFS, above the -40 dB threshold) it removed only 1.87%; at 5 dB it removed 0.65%.
- Cleaning the noise first fixes that: after the noise remover's Standard setting the same noisy files lost 34.6%, 36.7%, 34.2%.
- The threshold is fixed at -40 dB and the tool removes every silence, not only long ones. Natural pauses between sentences go too, so speech can sound rushed.
- It is fast: 0.19 s for the 30 second recording. The output is an MP3 at LAME quality 2. Files up to 100 MB are accepted.
What the tool does to your file
It runs ffmpeg's silenceremove filter: it trims leading silence, then removes every later stretch that stays below the threshold. A stretch of sound counts as silence when its level is under -40 dB, measured against digital full scale.
| Setting | Value | What it means |
|---|---|---|
| start_periods | 1 | Silence at the start is removed once |
| start_threshold | -40 dB | Quieter than this counts as silence at the start |
| stop_periods | -1 | Every later silence is removed, not just the first |
| stop_threshold | -40 dB | Quieter than this counts as silence |
| Output | MP3, LAME quality 2 | Saved as MP3 |
Measured: clean speech and speech with a fan
Measured on 30 seconds of a LibriVox reading of The Heart of a Mystery (public domain), as published (128 kbps mono MP3) and as a lossless WAV decode of the same recording. The tool's own ffmpeg arguments ran on the same single-thread WebAssembly engine your browser downloads; times are from a headless run and your device will differ.
The speech recording's average level is -23.1 dBFS. The noisy files add a real fan recording at 15 dB and 5 dB below that level (see the noise remover page), so the fan sits near -38 and -28 dBFS in the pauses.
| Recording | Original | After | Removed | Output size | Time |
|---|---|---|---|---|---|
| speech with natural pauses | 30.0 s | 20.611 s | 9.39 s (31.30%) | 296,207 bytes | 0.19 s |
| the same speech with fan noise at 15 dB | 30.0 s | 29.440 s | 0.56 s (1.87%) | 357,205 bytes | 0.24 s |
| the same speech with fan noise at 5 dB | 30.0 s | 29.806 s | 0.19 s (0.65%) | 347,018 bytes | 0.24 s |
Why noise defeats a silence remover
The filter only knows levels. A fan, an air conditioner or room tone sitting at -38 dBFS is louder than the -40 dB threshold, so every pause counts as sound and nothing is removed. In our test the 15 dB fan file lost 1.87% and the 5 dB file lost 0.65%, against 31.3% for the clean recording.
The tool cannot be told a different threshold. If your recording has a noise floor, clean it first.
Cleaning the noise first
We ran the three noisy files through the noise remover on its Standard setting and then through this tool. The pauses were now quiet enough to count as silence.
| Recording | Original | After | Removed |
|---|---|---|---|
| fan at 15 dB, cleaned with the noise remover (Standard) | 30.0 s | 19.608 s | 10.39 s (34.6%) |
| fan at 5 dB, cleaned with the noise remover (Standard) | 30.0 s | 18.984 s | 11.02 s (36.7%) |
| street at 5 dB, cleaned with the noise remover (Standard) | 30.0 s | 19.752 s | 10.25 s (34.2%) |
That is the clean-recording result again: a third of the file.
Free tools:Remove background noise from audioNormalize audio loudnessCut an audio fileRemove background noise from video
When to use it, and when not
- Voice memos, lecture recordings and dictation, where long gaps are dead air.
- Cutting the silence from the start and end of a take.
- Not for conversation or interviews you want to sound natural: removing every pause makes people talk over each other.
- Not for music: quiet passages and fades fall below -40 dB and are cut.
Questions people ask
How do I remove silence from an audio file?
Choose the file and run the tool. It removes every stretch quieter than -40 dB. On our 30 second speech recording it removed 9.4 seconds.
Why did it remove almost nothing?
Background noise above -40 dB is not silence. Our recordings with a fan added at 15 dB and 5 dB below the voice lost only 1.9% and 0.7%. Remove the noise first.
Can I change the threshold or keep short pauses?
No. The threshold is fixed at -40 dB and every silence is removed.
Will the speech sound rushed?
It can. About a third of our test recording was pauses, and all of them were removed, including the ones between sentences.
What format do I get?
An MP3 at LAME quality 2. Files up to 100 MB are accepted.
Is my audio uploaded?
No. It is processed in your browser.
Related
Sources and how we checked
- FFmpeg Filters Documentation (read 2026-10-06)
- FFmpeg Documentation: ffmpeg (read 2026-10-09)
Reviewed October 9, 2026. The measurements ran the tool’s own ffmpeg arguments on the same single-thread WebAssembly engine your browser downloads. The script, the results and the saved documentation are in data/seo/hub/evidence/2026-10-09/tools.