Donaldson R F Remove Silence from Audio
Trim the dead air from a recording. You set how quiet counts as silence and how long a gap must last; the waveform shows exactly what will be cut.
How silence is detected
The recording is examined in slices of 10 milliseconds. A slice counts as silent when its average level (RMS, both channels together) is at or below the level you choose, measured in decibels relative to the loudest possible sound: −40 dB is a good start for a quiet room, −30 dB for a noisy one, and −55 dB or lower for very clean recordings. A run of silent slices is removed only if it lasts at least the minimum length you set, so natural pauses between words that are shorter than that stay untouched. Around each cut the tool keeps a little silence (0.1 s by default) so speech does not sound clipped, and it joins the pieces with a 5 ms cross-fade.
The shaded parts of the waveform update as you move the settings, and the summary tells you how much will be removed. If a breath or a soft word is being cut, lower the level (for example from −40 to −50 dB); if pauses are not being removed, raise it or shorten the minimum length. Silence at the very start and end is trimmed too (leaving only the small amount you chose to keep); choose Only at the start and the end to leave the pauses inside the recording alone. Related tools: the MP3 cutter and volume changer (raise a quiet recording first if its speech is close to the silence level).
Frequently asked questions
What level should I use?
Start at −40 dB. The waveform preview shows what will be cut, so you can adjust before saving.
Will it cut into my words?
Only if the level is set too high or the shortest-silence length too small. Keep some silence around speech (default 0.1 s) to leave a natural pause.
Is my file uploaded?
No. It is analysed and edited in your browser.