Remove Podcast Pauses Without Cutting Quiet Speech
Long gaps can slow an episode down. Soft words and intentional pauses belong in it. Review what a silence tool proposes before shortening the conversation.
By the fix.fm team · Published · Updated
A shorter episode should still sound like a conversation
Shorten a pause when it makes the listener wait for no useful reason. Keep it when it gives a thought room to land, separates speakers, or lets someone take a natural breath. A faster episode is not automatically a clearer episode. Listen to a few sentences around the edit, rather than judging the quiet section on its own.
The silence trimmer proposes cuts by level and duration. It cannot tell whether someone is thinking, whispering, or waiting for an answer. Those decisions stay with you. Start on a copy of the recording and keep the unedited source. If a video or a transcript is already synchronized to it, deleting time will move everything after each cut.
What counts as silence here
Digital silence is a sequence of zero samples. A quiet recording is usually different: room tone, breathing and distant noise remain between words. Our tool calls a block quiet when its RMS level is below the threshold you choose. Each block is about ten milliseconds long, and every channel must qualify at the same time.
That distinction is why a stereo conversation needs linked edits. A pause on the left may contain speech on the right. Cutting each channel independently would shift their timing and could remove the other person's answer. The Audacity Truncate Silence manual describes the same important choice between synchronized and independent edits, and warns that low-level fades can also qualify as silence. Our tool always links the channels.
Set the threshold, minimum and retained gap separately
These settings answer different questions. The threshold decides how quiet a passage must be. The minimum decides how long it must remain quiet. The retained gap decides how much breathing room remains after the cut. Raising the threshold means making it less negative, which admits louder passages into the candidate list.
| Setting | Conservative starting point here | What to listen for |
|---|---|---|
| Quiet threshold | -45 dBFS RMS | Soft words and breaths must stay above it, or be manually kept |
| Minimum gap | 1 second | Brief breaths should not become candidates |
| Retained gap | 0.25 seconds | The next sentence must not feel abrupt |
These are our tool defaults, not a podcast delivery standard. In a noisy room, -45 dBFS might find no gaps at all. Increasing it may reveal gaps, but can also include a quiet speaker. Lowering it protects softer material. If the whole file falls below the threshold, the tool proposes no cuts; try a lower threshold before assuming the recording is empty.
Here is a repeatable timing example. Put one second of tone on each side of a two-second silent gap. With a one-second minimum and 0.25 seconds retained, the middle gap loses 1.75 seconds. The four-second file becomes 2.25 seconds. The tool leaves 0.125 seconds on each side of the join. Keeping that proposed cut leaves the full four seconds instead. Our fixture checks also insert a known gap into a public-domain spoken recording and verify that the surrounding voice samples remain intact.
Review before applying
- Choose the recording and enter conservative settings.
- Find quiet gaps. Check the proposed removal times and total time removed.
- Preview each cut's original context. Untick gaps that contain a soft word, a meaningful pause, a breath you need, or another person's response. Review subsequent pages if there are many cuts.
- Apply the selected cuts, then play the edited audio. Listen for rushed transitions and compare with the original in your editor if a join feels wrong.
- Return to the cut list to keep more pauses or change the settings. Save a separate WAV editing copy when you are satisfied.
Optional five-millisecond fades take the two sides of each join toward zero to soften a boundary click. They do not overlap the segments or change the length further. They also cannot hide a cut through a word. If the edit feels wrong, keeping more of the pause is usually more useful than adding a stronger fade.
Shortening pauses does not denoise a voice
This edit removes time. A fan that continues beneath speech will still be present after the gaps are shortened. Use background noise repair for steady hiss, or the repair finder to distinguish hum, wind and room echo. Then review the cuts again: processing can change which passages fall below the threshold.
After editing, use the LUFS meter and listen to the quietest speaker. Integrated loudness uses gates, so it does not simply average every silent sample into the reading, but changes to content and pacing can still change the result. Choose the final format using WAV vs MP3; an MP3 re-encode can add padding, while WAV keeps the edited sample count.
For the complete order of recording, editing, cleanup and export, return to the podcast audio workflow. A level-based silence tool is useful for finding candidates. A timeline editor remains the better place for detailed storytelling edits and synchronized audio and video.