Silence Remover

How the Silence Remover works

Silence remover: how it works

Speech is full of gaps. Thinking pauses, page turns, sips of water, the silence while someone finds the next slide. A raw hour of interview or lecture typically has several minutes of dead air in it, and listeners feel every second. This page finds those gaps and tightens them, leaving the words untouched and the pacing natural.

Three settings. The threshold decides what counts as silence, in decibels below full scale; -45 dBFS works for most rooms and should be raised if the recording is noisy. The minimum length decides which pauses are worth cutting — anything shorter is left alone, so the natural rhythm between words and sentences survives. The default of 700 ms catches thinking pauses but not ordinary sentence breaks. The keep value is what each long pause is shortened to; 250 ms leaves a clear breath between phrases, while zero butts them together, which sounds rushed for speech but is right for chopping up sound-effect takes.

How the cut is made. The file is scanned in ten-millisecond windows and each window is judged loud or quiet. Runs of quiet windows longer than the minimum are shortened to the keep length, taking the silence from the middle of the gap so the tail of one word and the start of the next are preserved. A five-millisecond fade is applied at each side of every join, which is enough to prevent clicks without being audible.

Checking the result. The stats panel shows the number of pauses found, the total time removed and the new length before you export anything, and Preview plays the tightened audio. If the count is suspiciously high, the threshold is probably catching quiet speech; if it is zero, the room noise is above the threshold and needs a higher value.

What it is not. This is not a noise gate and does not change levels during speech, and it is not a de-noiser. It removes time, nothing else. For a recording with an audible hum, the hum will still be there in the shortened gaps.

Related tools