# Auto captions and subtitles

Typing and timing captions by hand eats hours on every talking video

These tools transcribe your audio and place timed captions for you, either burned into the video or as editable text in your editor. You still read them over before export

## See it work

Visuals (real captures, with sources):

* Clip: Before and after: animated captions with a word by word highlight (README of kajisho5/ffmpeg-skill: https://github.com/kajisho5/ffmpeg-skill#readme) https://agentcut.dev/media/ffs-captions.mp4
* Clip: A vertical clip with burned in captions, from the podcli README (README of nmbrthirteen/podcli: https://github.com/nmbrthirteen/podcli#readme) https://agentcut.dev/media/podcli-demo.mp4
* Video: MrRaoAI has Claude Code cut and caption Reels (MrRaoAI on YouTube: https://www.youtube.com/watch?v=NnDAgw3lfX4)
* Video: Yash builds an auto caption setup with Claude Code instead of paying for Opus Pro (Yash | AI Automation  on YouTube: https://www.youtube.com/watch?v=8YA-eC9n5L4)
* Video: Jason Cooperson sets up the Premiere Pro MCP Bridge and lets Claude edit in Premiere (Jason Cooperson on YouTube: https://www.youtube.com/watch?v=Ohyo6-VO1jY)

## Real examples

* The maker of podcli renders vertical clips for the Deeptech Decoded podcast with it and fixes what shows up in those real renders, like the crop landing on a wall when a clip switches from full screen to split screen (nmbrthirteen, the maker of podcli, GitHub pull request 2026-10-01: https://github.com/nmbrthirteen/podcli/pull/242)
* Uses ffmpeg-skill on talking head videos with a webcam or screen recording picture in picture, which also get reframed, captioned, loudness normalized and exported for a platform, and asked for that whole job to fit in one project file (rymalia, GitHub user, GitHub issue 2026-10-12: https://github.com/kajisho5/ffmpeg-skill/issues/326)
* Had Codex drive Jianying, the Chinese version of CapCut, through VectCutAPI to mass produce quote videos for Douyin. Once the first template worked, 7 more videos took 4 minutes, and each new one takes under 2 minutes. Anything off gets fixed by hand in the draft (@Sixtimenight, creator, in Chinese, jxxy.net 2026-08-01: https://www.jxxy.net/ai/articles/sixtimenight-codex-vectcutapi-quotes-video/)
* Connected Claude to Premiere Pro so the AI cuts silences and flubs, adds synced captions and music, and a person does the final check. They delivered 4 short dramas to paying clients this way (Macky, runs a video company in Japan, in Japanese, note 2026-08-13: https://note.com/macky_creators/n/n5940c91cd0b4)
* Used Codex and FFmpeg to turn two How I AI recordings into 8 captioned vertical Shorts, then had it upload them in YouTube Studio after she approved each one (Claire Vo, host of How I AI, clairevo.com 2026-09-22: https://clairevo.com/ai-workflows/edit-and-publish-youtube-shorts-with-codex-and-ffmpeg)
  > The biggest improvement came from giving it an actual editing brief

## Ideas to try (ideas, not user reports)

* Burn captions into every Short in a folder with one request, then check the first one before running the rest
* Ask the agent to keep captions away from faces and screen content, the way you would brief a junior editor

## Tools

* [ffmpeg-skill](https://agentcut.dev/tools/ffmpeg-skill/): Captions, 9:16 crops, silence removal and exports done on your own computer by an AI. Works with FFmpeg, easy setup, free
* [VectCutAPI (CapCut/Jianying API)](https://agentcut.dev/tools/vectcutapi-capcut-jianying-api/): Making lots of CapCut videos with subtitles, voiceover and effects without editing each one by hand. Works with CapCut, medium setup, freemium
* [MCP Bridge for Adobe Premiere Pro](https://agentcut.dev/tools/mcp-bridge-for-adobe-premiere-pro/): Cutting pauses, building rough cuts and doing timeline busywork in Premiere by asking an AI. Works with Premiere Pro, medium setup, free
* [podcli](https://agentcut.dev/tools/podcli/): Turning a long podcast episode into vertical clips with face tracking and captions. Works with FFmpeg, Premiere Pro, Final Cut Pro, easy setup, free
* [PremierPro MCP Server (ayushozha)](https://agentcut.dev/tools/premierpro-mcp-server-ayushozha/): Handling Premiere timelines, audio, captions, graphics templates and exports from an AI chat. Works with Premiere Pro, hard setup, free
* [Premiere Pro MCP Server (170+ tools)](https://agentcut.dev/tools/premiere-pro-mcp-server-170-tools/): Editing, adding effects and captions, and exporting in Premiere from Claude. Works with Premiere Pro, medium setup, free
* [Kinocut (formerly mcp-video)](https://agentcut.dev/tools/kinocut-formerly-mcp-video/): Repurposing long videos into Shorts and Reels with quality checks before export. Works with FFmpeg, easy setup, free
* [DaVinci Resolve MCP Bridge (Free-version)](https://agentcut.dev/tools/davinci-resolve-mcp-bridge-free-version/): Using AI with Resolve Free, including local transcription, without paying for Studio. Works with DaVinci Resolve, medium setup, free
* [CapCut MCP Server — Extended](https://agentcut.dev/tools/capcut-mcp-server-extended/): Making talking head CapCut edits with styled captions and animation presets. Works with CapCut, hard setup, free
* [cutcli cookbook](https://agentcut.dev/tools/cutcli-cookbook/): Turning a prompt or template into an editable CapCut draft. Works with CapCut, easy setup, freemium
* [HyperFrames (HeyGen)](https://agentcut.dev/tools/hyperframes-heygen/): Making motion graphics, captions and callouts with AI and rendering them to MP4. Works with FFmpeg, Programmatic video, medium setup, free

---
AgentCut: https://agentcut.dev/ · All tools as JSON: https://agentcut.dev/tools.json · Overview for agents: https://agentcut.dev/llms.txt
