--- title: "Remove Pauses" description: "Cut the pauses out of a talking-head take: a jump cut on every breath, picture and sound cut together so lip sync holds. Wire the take, with its audio, into `video`. It finds the pauses on the take's own audio (no transcript, no paid model) and removes every one longer than `min_pause_ms` (default 150), keeping `pad_ms` (default 40) of each pause on both sides so word edges are not clipped, with a 5 ms audio fade at every cut so there are no clicks. Silence before the first word and after the last is cut too. `noise_db` (default -35) is how quiet counts as a pause: lower it (for example -45) if quiet words get cut, raise it (for example -30) in a noisy room. Optional `speed` (1 to 1.15, default 1) speeds the cut up afterwards with the pitch kept. Out-of-range values are clamped, and the values used come back in the cut list. Outputs: the cut `video`, and `cuts`: every kept segment with its source and output times, how many pauses were cut, and the length before and after, for captions and an audit. The video is re-encoded (x264, CRF 18); a take with nothing to cut at 1x comes back unchanged. Takes up to 180 seconds. Free to run: this node quotes and charges 0 credits." updated: 2026-10-31 --- # Remove Pauses **Node type:** `utility:remove_pauses` **Category:** `Editing` ## Description Cut the pauses out of a talking-head take: a jump cut on every breath, picture and sound cut together so lip sync holds. Wire the take, with its audio, into `video`. It finds the pauses on the take's own audio (no transcript, no paid model) and removes every one longer than `min_pause_ms` (default 150), keeping `pad_ms` (default 40) of each pause on both sides so word edges are not clipped, with a 5 ms audio fade at every cut so there are no clicks. Silence before the first word and after the last is cut too. `noise_db` (default -35) is how quiet counts as a pause: lower it (for example -45) if quiet words get cut, raise it (for example -30) in a noisy room. Optional `speed` (1 to 1.15, default 1) speeds the cut up afterwards with the pitch kept. Out-of-range values are clamped, and the values used come back in the cut list. Outputs: the cut `video`, and `cuts`: every kept segment with its source and output times, how many pauses were cut, and the length before and after, for captions and an audit. The video is re-encoded (x264, CRF 18); a take with nothing to cut at 1x comes back unchanged. Takes up to 180 seconds. Free to run: this node quotes and charges 0 credits. ## Canvas ports These appear as port handles on the left side of the node. | ID | Label | Details | | ------- | --------------------------- | -------------------- | | `video` | **Video (take with audio)** | `VIDEO` _(required)_ | ## Sidebar config These render as form fields in the right-side config panel when the node is selected. | ID | Label | Details | | -------------- | --------------------------------------- | ------------------------- | | `min_pause_ms` | **Min pause to cut (ms, 50 to 5000)** | `NUMBER` · default: `150` | | `pad_ms` | **Pause kept each side (ms, 0 to 500)** | `NUMBER` · default: `40` | | `noise_db` | **Noise floor (dB, -90 to -10)** | `NUMBER` · default: `-35` | | `speed` | **Speed after the cut (1 to 1.15)** | `NUMBER` · default: `1` | ## Outputs | ID | Label | Type | | ------- | ------------- | ------- | | `video` | **Cut Video** | `VIDEO` | | `cuts` | **Cut list** | `JSON` | ## What it is for Fast talking-head reels cut every breath. A generated or recorded take leaves a beat after almost every sentence, and those beats add up: a take that is one fifth pauses plays at about 150 words a minute, where a reel that holds attention runs well over 200. Remove Pauses is that jump cut, done for you. Wire the take into **Video** and run it. It listens to the take's own audio, finds every pause, and cuts it out of the picture and the sound at the same frames, so the lips stay in sync across every cut. ## Recipe: a tight talker 1. Talking-head take (a talking-head video node, or an imported clip) into **Video**. 2. Run it with the defaults and watch the result. Read the summary on the card: how many pauses were cut, and the length before and after. 3. Words clipped at the edges? Raise **Pause kept each side** (60 to 80 ms). Quiet words cut out? Lower **Noise floor** to -45. Room noise keeping pauses in? Raise it to -30. 4. Still slow? Set **Speed after the cut** to 1.05 or 1.1. The pitch is kept. 5. Wire **Cut Video** on into captions and the Video Editor. Caption timing should come from the cut video, not the original take. ## The dials - **Min pause to cut** (default 150 ms): only quiet spans at least this long are cut. Lower cuts more, higher leaves short breaths in. - **Pause kept each side** (default 40 ms): how much of each pause stays on either side of the cut, so the end of one word and the start of the next are not clipped. Each cut also gets a 5 ms audio fade, so there are no clicks. - **Noise floor** (default -35 dB): anything quieter counts as a pause. - **Speed after the cut** (1 to 1.15, default 1): speeds the whole cut up. Values out of range are clamped, and the values used come back in the cut list. Silence before the first word and after the last is cut too. ## Outputs - **Cut Video**: the take with the pauses removed, same size and frame rate. - **Cut list**: every kept segment with its start and end in the original take and in the cut video, how many pauses were cut, and the length before and after. Use it to move caption or overlay timing from the original take onto the cut. ## Limits Takes up to **180 seconds**, and at most 150 kept segments in one run (raise Min pause or trim the take for more). The take needs an audio track. The video is re-encoded (near visually lossless); a take with nothing to cut at 1x comes back unchanged. This node is free: it quotes and charges **0 credits**. --- _Auto-generated from the Wireflow node registry._ Last updated: 2026-10-31 --- Documentation index: fetch https://www.wireflow.ai/llms.txt for the full list of Wireflow docs.