Smart Clipping
Find the silence and dead air in a take automatically, review the proposed cuts, then apply them in one go.
Smart Clipping listens through a recording for stretches where nothing is happening — silence, dead air, you reaching for the mouse — and proposes cuts. You review the list and apply it.
It proposes. It never cuts on its own.
Available on every plan, including Free.
Presets
Start with a preset, then fine-tune if the result is close but not right.
| Preset | Counts as silence below | Must last at least |
|---|---|---|
| Conservative | −45 dB | 3.0s |
| Balanced | −40 dB | 2.0s |
| Aggressive | −35 dB | 1.0s |
Balanced suits most narrated demos. Conservative is the one to reach for when you speak with long deliberate pauses and don't want them removed — a pause you meant is indistinguishable from dead air by volume alone, only by length. Aggressive is for an unscripted take you know is loose.
Fine-tuning
| Control | Range | Does |
|---|---|---|
| Silence threshold | −60 dB to −20 dB | The level below which audio counts as silence |
| Min silence length | 0.3s to 5.0s | Gaps shorter than this are left alone |
| Edge padding | 0 to 1000 ms | How much audio to keep either side of speech |
Edge padding is the one that decides whether the result sounds natural. Cutting at the exact moment speech stops clips the tail off words and makes consecutive cuts sound abrupt. Leaving a little room either side keeps the rhythm of the sentence.
If a cut sounds clipped, raise the padding before touching anything else.
Why volume alone isn't enough
A quiet moment isn't always a dead one. You might be silent because you're dragging a window, selecting text, or waiting for something to load — all of which the viewer needs to see.
Smart Clipping reads the cursor and keystroke data captured alongside the audio and correlates it against the silence, so a quiet stretch where you were clearly still working is treated differently from a quiet stretch where nothing happened at all.
That correlation needs the cursor track, which needs the Accessibility permission at the time of recording. Without it, Smart Clipping falls back to audio alone and will propose cutting quiet stretches where you were working silently. See Permissions.
Reviewing
Every proposed cut is listed, and you can jump the playhead to any of them to hear it in place before deciding. Listen to the ones at the start and end of your narration first — those are the cuts most likely to clip a word.
Applying writes the cuts to the timeline as ordinary edits, so ⌘Z undoes them and you can adjust any one of them afterwards by hand.