Load a recording, set the silence threshold in dBFS and the shortest gap worth removing, then press Remove Silence. Review the time saved and download the shortened WAV.
The audio is scanned in ten millisecond windows and the RMS level of each window is compared with your threshold in dBFS. Runs of quiet windows longer than your minimum gap are removed, and everything else is kept.
Start around -45 dBFS for a clean studio recording and raise it towards -35 dBFS for noisier rooms. If words are being clipped off, lower the threshold; if pauses survive, raise it. The report tells you exactly how much was removed each time.
Cutting exactly at the threshold chops the natural attack and tail of words and sounds unnatural. The tool keeps a short pad on both sides of each kept region so speech still breathes while long silences disappear.