What this does
Scans your audio track for silent gaps exceeding a specified threshold and instantly ripple-deletes them across all timeline tracks.
It removes gaps between spoken words, so it is the right tool for pauses, breaths and filler on a talking head or interview. It is the wrong tool on a screen recording, demo or gameplay capture: that footage is largely unnarrated, so an audio-driven pass can read most of it as removable. For those, describe what you want cut visually — 'cut the part where the page is loading' — instead.
How to do it
Follow these step-by-step instructions in Cubix.
Open your project with spoken audio
Ensure your video has an active speech audio track selected on the timeline.
Raw footage timeline before silence removal.
Ask Cubix to remove silences
Type your request in the AI Agent box or click the 'Remove Silences' Quick Action pill.
Compare raw unedited video pauses vs. Cubix silence removal.
Choose how tight it should feel
Rather than guessing a threshold in milliseconds, describe the feel you want. Gentle preserves thinking pauses and suits interviews, podcasts and long-form. Medium is the default for most talking-head content. Aggressive produces the gapless jump-cut rhythm that short-form expects.
Processed timeline without silences.
Check what was removed
Cubix reports what went — the amount removed, the number of cuts, and the timestamps of any long stretches it took out. Read that before moving on, because a long silent section is a different thing from a pause.
Decide what else the pass should do
Beyond the pacing mode, the pass can snap its cuts to a musical beat, add speed emphasis, smooth the joins between lines of dialogue, stabilise the picture, and drop chapter markers at the section boundaries it finds. Open its card to switch any of those on.
Hide the visible jumps
Each removed pause leaves a small discontinuity on a locked-off camera. A slight scale change across the cut, or a cutaway over the busiest run of them, is what turns a sequence of jolts into something that reads as deliberate.
Pro Tips for Best Results
- Cuts are placed on word boundaries rather than on a raw level threshold, which is why the result still sounds like speech instead of a splice.
- A long unnarrated stretch counts as silence and will be removed along with the pauses. If you want one kept, say so: 'tighten it but leave the quiet section where I am demonstrating the tool'.
- The whole pass is a single undo, so it costs nothing to try the aggressive version and step back from it.
- Ask for a specific pause to be restored — 'put back the pause after I ask the question' — rather than undoing the entire pass to recover one beat.
- Maximize Retention is the heavier relative of this pass: as well as cutting, it punches in, adds sound hits, a hook and captions. Its hold setting runs Tight for talking heads, Balanced to keep a beat of every shot, and Roomy for vlog and B-roll footage.
- Beat snapping, speed emphasis, dialogue transitions, stabilisation and chapter markers are all optional extras on this pass — open its card if you want any of them.
Removing pauses and filler words does not touch a single word you actually said, which is why it is usually the first edit worth making on any spoken video.
Frequently Asked Questions
Common questions around this editing workflow.