How to Remove Long Pauses From Podcast Video

A recorded conversation is full of air: the wait while a guest finds their words, the gap while someone checks a note, the half-minute before anyone remembers the camera is rolling. This post is about how to remove long pauses from podcast video without clipping words or turning a conversation into a breathless list of sentences. It covers which pauses to cut, which to keep, what silence removal cannot do, and the settings that get the result in one pass.

Why more podcasts are being edited as video

This is not breaking news, but it is why the question keeps coming up. On 17 September, Spotify announced that its Partner Program for video podcasts is coming to more than 35 new markets later this autumn, including Brazil, Mexico, Italy and Spain. Spotify says video shows in the programme have seen "more than 45% average growth in consumption hours since its launch", a figure Podcast News Daily reported the same day.

When a show is audio only, a long pause is a small cost. On video, the viewer is looking at two people not talking, and that is when they reach for the next video. Dead air costs more on screen than it ever did in an earbud.

Pauses worth cutting and pauses worth keeping

The goal is not zero silence. A conversation with every gap removed sounds like a recording played too fast, and it loses the rhythm that makes two people talking worth watching. Sort pauses into three kinds.

• Dead air. The setup before the conversation starts, a guest searching for a file, a long stall while someone loses the thread. Cut all of it.
• Long gaps inside an answer. A speaker stopping to think for three or four seconds. Usually worth shortening, because on video it reads as hesitation, not thought.
• Conversational beats. The short pause between one person finishing and the other starting, the breath before an answer, the beat after a joke lands. Keep these. They are the difference between a conversation and a transcript read aloud.

The settings below are how you tell a tool to take the first two and leave the third.

What silence removal can and cannot do

Silence removal in flshy reads the audio level, not the words. Anything quieter than a threshold for long enough counts as silence. That makes it fast and predictable, and it sets clear limits:

• It does not hear filler words. "Um", "uh", "you know" and false starts are speech as far as the audio level is concerned, so they stay. Removing them needs a tool that works from a transcript, or a manual cut.
• It needs the gaps to be quiet. If a music bed runs under the conversation, the pauses are not silent and will not be found. Cut the silences before you add music, not after.
• It cannot pick a speaker. Crosstalk and laughter are sound, so they count as speech. That is usually what you want.
• A cut edge can land close to a word. Because it never reads a transcript, it has no idea where a word begins. Padding is the control that keeps the edge clear.

The settings that matter for a podcast

The panel has four sliders, and three decisions.

Threshold: what counts as silence

The threshold is a level in decibels, from -80 to -20, and -40 is the default. Anything quieter than it, for long enough, is treated as silence, so raising it treats more of the recording as silent.

Start at the default. A quiet, treated room will usually be fine there. A noisy room needs a higher threshold, because the room's own noise floor is louder. If fans, traffic or a laptop keep the gaps from ever dropping below the threshold, the tool finds nothing to cut. Move the threshold up until the background counts as silence and speech does not.

Minimum gap: what counts as a long pause

The minimum length a gap has to reach before it counts is the control that protects the conversation. Set it short and every breath between sentences becomes a cut. Set it long enough that the ordinary beat between speakers survives, and only the real dead air goes. For a conversation, err on the long side the first time; you can always run it again tighter.

Padding: keeping words whole

Pad Before and Pad After keep a little audio either side of every cut. Both default to 0.1 seconds and go up to 2 seconds. If speech sounds clipped at the start or end of a sentence after cutting, raise them. For a relaxed interview, a little more padding keeps the breath before an answer, which is most of what makes it sound like a person talking.

Removing long pauses in flshy, step by step

All of this runs free in flshy, in the browser. The audio is analysed and cut on your own machine, and nothing about this feature touches the network. If keeping footage off other people's servers is part of why you are here, see a free CapCut alternative that keeps footage local. The reference for every control is remove silence from a recording.

1. Bring the recording in. Drag the episode onto the timeline. It works on video and audio clips alike.
2. Open Remove silent gaps. Click the clip, open the Audio tab in the inspector, and find the section called Remove silent gaps.
3. Set the three decisions. Threshold for your room, minimum gap long enough to keep the conversational beats, and padding that keeps words whole.
4. Analyse, and read the proposal. Nothing moves yet. You get the number of silent sections found, the total time they would remove and what percentage of the clip that is, the length the clip would end up, and the first eight gaps listed by timecode.
5. Sanity-check the numbers. If it would remove a large share of a conversation, the threshold is probably catching quiet speech: lower it, or lengthen the minimum gap. If it found nothing, the panel says so and tells you which way to move the sliders.
6. Cut. The button states the whole edit before you press it, in the form Remove 9 silences, saves 12.4s. One press applies it, and it is a single step in the undo history.
7. Listen to the joins. Play through a few of the cuts. If one sounds abrupt, undo, raise the padding a touch and cut again.

Then cut the clips

A tightened episode is also the best raw material for short clips, because every moment you pick already starts on the speech. Once the long version is done, the order that avoids repeating work is captions after cutting, then reframing:

• Caption cues sit on their own track and are not moved by a silence cut, so add captions to the clip once it is at its final length.
• A two-person podcast shot wide will not survive a vertical crop with both faces in it. Our post on turning horizontal video into full-screen Reels covers what to do instead, and turning long videos into original YouTube Shorts covers what YouTube's October update means for clips cut from your own episodes.
• The reframing itself is in turn a landscape video vertical.

Export is H.264 MP4 up to 4K with no watermark, and no length limit on the episode.