lcbuaaliu/ai-jian-koubo — explained in plain English
Analysis updated 2026-05-18
Automatically flag filler words, hesitations, and long silences in a recorded video for removal.
Review and adjust suggested cuts in a local browser interface synced to the video waveform and transcript.
Export an FCPXML project file that drops directly into Final Cut Pro or DaVinci Resolve.
Generate a clean, timestamp-free subtitle file from a video's audio track.
| lcbuaaliu/ai-jian-koubo | feigaobox10/ai-economics-reader | akisato57/aki-bangumi-vault | |
|---|---|---|---|
| Stars | 29 | 29 | 30 |
| Language | HTML | HTML | HTML |
| Setup difficulty | moderate | easy | easy |
| Complexity | 3/5 | 1/5 | 2/5 |
| Audience | vibe coder | general | general |
Figures from each repo's GitHub metadata at analysis time.
Requires Node.js, Python 3, ffmpeg, curl, and a free Volcengine API key for transcription.
AI Jian Koubo is a skill for AI coding agents that automates the tedious part of editing talking-head videos. The time-consuming step when editing a podcast or presentation recording is not the actual cutting, but figuring out which moments to remove: repeated phrases, hesitations, verbal tics, and long silences. This tool handles that identification step automatically and hands a project file to a video editor to finish the job. The workflow has two modes. In the first mode, the agent takes a video file, extracts the audio, sends it to a transcription service (Volcengine, a Chinese cloud platform from ByteDance), reads the full transcript, and flags the segments that look like mistakes or filler. It then opens a local review webpage in your browser showing a waveform with the transcript below it. You can click words in the transcript to mark them for removal, adjust the silence padding, and see exactly which video frame each cut would land on. When satisfied, you click an export button to generate an FCPXML file, which is a standard project format readable by both Final Cut Pro and DaVinci Resolve. You drag that file into your editing software to apply the cuts. In the second mode, the agent just produces a clean subtitle text file from the video with no timestamps, formatted for direct reuse. Everything runs locally on your machine. The video file is not uploaded to any third-party storage. The only external call is the audio going to Volcengine for transcription, which offers 40 hours of free usage, enough for a large volume of recordings. You need a Volcengine API key, which the README shows how to create in about two minutes. The skill is designed to be agent-agnostic: any coding agent that can read files, run shell commands, and follow a SKILL.md instruction file can use it. The repository describes installation as a single step: give the agent the repository URL and tell it to install, and it reads the README, sets up the environment, and runs a self-check automatically. Required tools on your machine are Node.js, Python 3, ffmpeg, and curl.
AI Jian Koubo is a skill that lets AI coding agents automatically find filler words and dead air in talking-head videos and export ready-to-use editing cuts.
Mainly HTML. The stack also includes Node.js, Python, ffmpeg.
The README does not state a specific license.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
Verify against the repo before relying on details.