Clean Captions: Trim Silences and Add Captions to Any Video Automatically
Drop a raw recording in, get back the same video cut tight and captioned word by word.
It takes a recorded video, cuts the silences and repeated takes, and burns in captions synced word by word, keeping the original orientation.
Any AI assistant that can read a SKILL.md file (Claude, Claude Code, Cursor, or Codex). No separate app, no account, no subscription, and no API key.
Works with
Clean Captions is an AI skill that does the two edits every talking-head video needs: it trims the silences and dead space, and it adds captions synced to your speech. You drop a raw recording into your AI assistant, the skill transcribes it with forced word-level alignment, cuts the pauses and repeated takes (keeping your last, cleanest take), and burns bold karaoke-style captions that light up word by word. It keeps the exact orientation you filmed in (vertical, horizontal, or square) and never crops or restyles your footage. If you have ever searched how to add captions to video or how to remove silence from a video without opening a full editor, this is the shortcut: one instruction in, one clean captioned file out.
- Add captions to any video automatically, synced word by word so viewers can watch with the sound off
- Remove silences and dead space so the whole clip feels tight and holds attention
- Cut repeated takes and fumbled lines, keeping only your last clean delivery
- Keep your original orientation and framing: no forced cropping to vertical, no restyling
- Skip the timeline editor entirely and finish a raw recording in one pass
- Get faithful captions of exactly what you said, with obvious transcription slips fixed, never rewritten
- 1Drop it into your AI assistant
Add the SKILL.md file to Claude, Claude Code, Cursor, or Codex. There is nothing to install and no account to create; your assistant reads the skill and knows how to run the whole flow.
- 2Give it your recording
Hand over a raw video file (or a link it can download). That is all it needs to start: one recording, no separate project setup.
- 3It transcribes and aligns the speech
The skill transcribes your audio and locks every word to its exact timestamp, so the captions land in perfect sync and the cuts fall between words instead of chopping them.
- 4It trims silences and repeated takes
It removes the pauses and dead air, then finds the spots where you restarted a sentence and keeps only your last clean take. It flags anything unusual and asks before making judgment calls.
- 5It burns in the captions
Bold, high-contrast captions appear in short two to three word chunks and reveal in time with your voice, placed cleanly in the lower third and sized to your video's height.
- 6You get the finished video
The output is the same video you filmed, in the same orientation, now cut tight and captioned, ready to post. A poster frame comes with it so you can preview at a glance.
- A cleaned, captioned MP4 in your original orientation
- Silences and dead space cut out for a tighter runtime
- Repeated takes and retakes removed, last clean take kept
- Word-by-word karaoke-style burned-in captions
- A poster frame image for quick preview
- Faithful captions of your actual words, with obvious errors corrected
Vertical, horizontal, or square
The skill detects the shape of your source and keeps it, so a landscape interview stays landscape and a phone clip stays vertical. Nothing is force-cropped.
Captions only, no cutting
Prefer to keep every pause? Ask it to add captions without trimming, and it will caption the clip as filmed.
Just the silence trim
Only want a tighter cut with no captions? It can remove the silences and repeated takes and hand back the clean footage on its own.
SKILL.md · README.txt · LICENSE.txt — Single-seat commercial use. Do not resell or redistribute the file. Compatible with Claude, Claude Code, Cursor, Codex.
What is the Clean Captions skill?
It is a SKILL.md file that teaches your AI assistant to take a recorded video, cut the silences and repeated takes, and burn in captions synced word by word. It is built for talking-head and social clips where you want a fast, clean edit without opening a video editor.
Does it work with Claude, Cursor, and Codex?
Yes. It runs in Claude, Claude Code, Cursor, and Codex. Any AI assistant that can read a SKILL.md file can use it.
Do I need to install anything or pay for a subscription?
No separate app, no account, no subscription, and no API key. You drop the SKILL.md into your assistant and give it a video. Your assistant handles the rest.
Will it crop my video or change the way it looks?
No. It keeps your original orientation and framing, whether that is vertical, horizontal, or square. It does not restyle your footage, add stock b-roll, or force a 9:16 crop.
Will the captions be accurate?
The captions reflect exactly what you said. The skill fixes obvious transcription slips but never rewrites your words or changes your voice, so the meaning stays yours.
Do I need editing or coding skills to use it?
No. You give it a plain instruction and a video file. There is no timeline, no keyframes, and no code to write. The skill asks before anything it needs and returns a finished file.
An AI skill is a small, reusable instruction file (a SKILL.md) that teaches an AI assistant to do one job well. Clean Captions is exactly that: a Claude skill / AI agent skill you drop into your own assistant and run on demand, with no app, no subscription and no lock-in. You own the file forever.
It runs in Claude, Claude Code, Cursor, Codex and any assistant that reads a SKILL.md or a system prompt. Because it is just text, it is fully portable, and if it needs something to run, clean captions asks you for it first instead of inventing.
Creators reach for it as a podcast clips AI and an AI podcast clip generator, to turn a podcast into Shorts and to pull podcast highlights automatically, finding the best podcast moments without scrubbing the whole episode.
- Download the file after checkout you get the SKILL.md plus a short README and your license.
- Add it to your AI drop SKILL.md into your skills folder (for example
.claude/skills/) or paste its contents as a custom instruction. - Run it ask your AI to use Clean Captions, give it what it asks for, and it does the work.
Get this skill plus every other one in a single agent bundle, ready to drop into your AI.
Ready to run Clean Captions?
Clean Captions
It takes a recorded video, cuts the silences and repeated takes, and burns in captions synced word by word, keeping the original orientation.
SKILL.md (plain text, nothing runs on your machine) plus a README and your license.Checkout opens soon. Get early access and I will send you the link the moment it is live.
✓
You are on the list.I will email you the moment it is live.
