AUD · Audio tools

Split Audio by Silence

Shareable link

Settings are written to the URL as you change them. Nothing differs from the defaults yet.

Cut on the gaps, not on the clock

Load one long recording and get back the pieces between its silences. A vinyl side becomes separate tracks. An hour of voice memos becomes one file per thought. A batch of takes recorded in a single pass becomes one file per take.

Detected pieces are drawn and numbered on the waveform before anything is written, so the settings can be judged by looking rather than by exporting and checking.

Threshold and shortest break

Threshold is the level below which audio counts as silence. Digital silence is far below anything a microphone produces: room tone usually sits between -60 and -40 dB, and vinyl surface noise higher still. Start at -45 dB and raise it until the gaps are found.

Shortest break is how long the quiet has to last before it becomes a split. This is the control that separates a pause from an ending. Speech pauses run to about a second; the gap between two tracks or two takes runs longer. The default of 1.5 seconds sits between them.

Padding and shortest piece

Padding extends each piece at both ends into the silence beside it. Without it, pieces start exactly on the first sound, which clips the front of a word and sounds abrupt. 200 ms is enough to breathe.

Shortest piece drops fragments. A cough between takes or a click on a record would otherwise be written as its own numbered file. Raising this to a few seconds keeps the output to real material.

Detection resolution

Silence is measured on the same peak envelope the waveform is drawn from, one value per bin rather than one per sample. That is why the pieces on screen are the pieces in the ZIP, and why the shortest useful break is measured in tenths of a second rather than in milliseconds.

Removing gaps instead of splitting on them

To close the pauses inside one file and keep it as a single file, the silence remover does that job with the same detection controls.

Frequently Asked Questions

It finds the silent gaps in one recording and writes the audio between them as separate files, delivered as a ZIP. It reads WAV, MP3, M4A, AAC, Ogg, Opus, FLAC, AIFF, and WebM audio, and every piece keeps the source format.

The splitter cuts on a clock, into equal parts or fixed lengths. This one cuts on content, wherever the recording goes quiet for long enough, so the pieces line up with takes or tracks instead of with a stopwatch.

The threshold is below the recording's noise floor. Tape hiss, room tone, and vinyl surface noise all sit above -60 dB, so raise the threshold until the gaps register.

The shortest break is set too low and normal pauses in speech are qualifying. Raise it to 1.5 or 2 seconds, which is longer than anyone pauses mid-thought and shorter than a real gap between takes.

Yes, deliberately. Each piece is extended by the padding at both ends, so nothing begins on the first syllable or ends on the last. It comes from the silence next to the piece, not from the neighbouring piece.

Up to 60. Each finished piece is held in memory while the archive is built, which is what sets the limit rather than anything about the encoder.

Yes. Each piece is decoded from the source and written again in the same format, so a lossy source goes through a second lossy pass. Load a WAV or FLAC where that matters.

Explore Our Tools

Browse all tools