Cut Silence by shash.builds
Cuts silences and umm or uh fillers from a face recording or voice-over with ffmpeg, keeping a 0.15 s pad. Free, local, no API key.
Free
Works with the AI Agents you already use
Claude Code
Cursor
Codex CLI
Muse
Dots+21 moreClaude CodeCursorCodex CLIMuseDots
Manus
Grok
Grok Bot+18 agentsCut Silence by shash.builds
Example session with this skill installed
Cut the silences and the umms out of face.mp4. Here are the word timings: face.words.json.
- Read your context and instructions
- Compiled the cut silence by
7.0 s to 3.5 s: 1 cut (2 silences, 1 filler). Wrote face_cut.mp4 and face_cut.mp4.cuts.json (every part kept and removed, with times). The original face.mp4 is unchanged.
Connects securely to your tools. The creator never sees your data.
About this skill
Cuts the silences and the "umm"s out of a face recording or voice-over, keeping 0.15 s of room around every word. One command, ffmpeg only, on your own computer.
Talking-head videos get tighter when the dead air goes. This skill does that pass for you.
It finds every pause longer than 0.4 seconds with ffmpeg's silence detector and cuts it, leaving 0.15 seconds of room around your words so nothing sounds chopped. Give it the word timings from the Transcribe tool and it also cuts "umm", "uh", "er" and "hmm", plus the gap around them.
What you get
- A new video (or audio file) with the cuts made, picture and sound in sync
- A cuts file listing every part kept and removed, with times, so you can check it or line up edits
- A dry-run mode that shows the cuts before anything is written
You control
| Option | What it does |
|---|---|
-o out.mp4 | output file (default face_cut.mp4); the original is never overwritten |
--pad 0.15 | seconds kept before and after speech |
--min-silence 0.4 | shortest pause that gets cut |
--threshold -35 | silence level in dB: -30 for a noisy room, -45 if quiet words get cut |
--words FILE | word timings from transcribe: filler words get cut too |
--fillers "um,uh" | which words count as fillers |
--dry-run | print the cut list and the new length, write nothing |
Before and after
In testing, a 7.0 s test clip ("Today we build a website." pause, "Umm.", pause, "It takes seven days.") came out at 3.5 s with the pauses and the umm gone, video and audio in sync. The test used a synthetic voice, no real person.
Requirements: ffmpeg. No Python packages. Free and MIT licensed. Part of the shash.builds Toolbox.
Captions made before the cut no longer line up; transcribe the cut file again. Quiet breaths can be cut if they fall below the threshold for longer than --min-silence. Filler removal needs word timings from a transcript (the Transcribe tool writes them). The output is re-encoded (H.264, CRF 18, AAC 192k).
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed today
- Free with an Agensi account
Requires
Permissions
File Scopes
Runs ffmpeg locally. No network access. Refuses to overwrite the original file.