Remove silence from audio

Automatically shorten long silent sections in voice recordings, podcasts, lectures, or interviews. The tool runs locally in your browser with FFmpeg/WASM and lets you control both the detection threshold and the minimum duration required before a quiet section is treated as removable silence.

Bethemesh uses FFmpeg’s silenceremove filter. The dB threshold determines how quiet a section must be before it counts as silence, while the minimum duration protects brief natural pauses. Detected silent periods are shortened while roughly 0.1 second is kept around joins, and the result is encoded as WAV. This is level-based detection rather than speech recognition, so very quiet music or ambience can also be affected if it falls below the threshold. Defaults are −40 dB and 0.5 seconds. Detection uses signal level, not transcription or breath recognition. The start uses a zero start-duration setting; the adjustable minimum duration applies to subsequent silence detection. A short initial pause and an internal pause therefore need not be treated identically. Up to roughly 0.1 seconds of silence is retained at joins. Quiet ambience may be shortened just like pauses in speech. Listen to full phrases and word endings, since the resulting duration cannot be predicted from threshold alone.