Skip to content

Remove Silence from Audio & Video

Remove silence from audio and video by detecting quiet RMS windows and applying one shared edit plan to sound and picture. Adjust threshold, minimum quiet duration and padding before export.

Found a problem, or something that could be better? We read every message and we fix things quickly.

Remove silence from audio and video by detecting quiet RMS windows and applying one shared edit plan to sound and picture. Adjust threshold, minimum quiet duration and padding before export.

The tool uses a local 8 kHz analysis copy and the existing ffmpeg.wasm engine, not AI speech recognition. Preview removed regions and retained ranges. Audio exports as PCM WAV; video re-encodes to H.264/AAC MP4, so quality and metadata may change.

How to use

  1. Choose a supported mono/stereo media file within the size and duration limits.
  2. Set the quiet threshold, minimum run and padding, then analyze.
  3. Review the cut plan, export and play the edited file before downloading.

Frequently asked questions

Is this speech recognition?

No. It measures RMS energy in 20 ms windows. Quiet speech or music can be classified as silence, so review the proposed cuts.

Does it keep video and audio cuts together?

Yes. The retained ranges drive both trim and atrim before concat. Unusual input timestamps still require playback review in the actual exported file.

What are the limits?

Up to 80 MB, ten minutes and 100 retained segments. Video is limited to 1920×1080 pixels; only first video/audio streams are exported, with mono or stereo input audio.

Can I export a completely quiet file?

If the settings remove everything, no empty media file is generated. Lower the threshold or increase minimum duration or padding and analyze again.

Detect first, then edit

RMS is evaluated across both analysis channels so activity in either prevents a quiet classification. Short runs below the minimum are preserved, and padding keeps context beside audible sections.

The waveform shows the RMS envelope and removed regions, while the table lists exact retained ranges. JSON preserves settings and the plan for inspection.

Local processing and export trade-offs

The approximately 32 MB media engine loads on first use. Analysis and encoding run in workers; the original file is not uploaded. The engine may remain cached for other media tools.

WAV is uncompressed PCM. MP4 is re-encoded rather than copied losslessly; extra tracks and source metadata are not preserved. Listen and watch for clipped speech or abrupt transitions before using the result.