threerocks/article2media / skills/video-transcript
video-transcript
Unverified视频转文字:把一条视频转成带时间轴的逐字稿文章(article.md + transcript/segments.md)。支持抖音(链接、短链、整段分享口令)、YouTube、B 站链接,本地视频 / 音频(mp4、mov、mp3、m4a、wav…),.srt / .vtt 字幕文件。文字来源依次是人工字幕、平台原语言自动字幕、本地 Whisper 识别;自动转简体、补标点、分段、专有名词提示与纠错。可以单独用,也是"视频转图文 / 视频转小红书"组合的第一步。Use when the user wants a video's subtitles, transcript or text (视频转文字、逐字稿、提取字幕、视频文案、听写这段录音), or gives a Douyin / YouTube / Bilibili link or a video/audio/subtitle file that another article2media skill should work from.
视频转文字(video-transcript)
路径约定:文中的
${CLAUDE_SKILL_DIR}指本 SKILL.md 所在的目录。宿主没有替换这个变量时(命令里原样出现${CLAUDE_SKILL_DIR}),把它换成这个目录的绝对路径再运行;../<名字>/指同级的另一个 skill 目录。
把视频变成一篇逐字稿文章 article.md:之后可以直接交付,也可以交给改写、翻译、长图、卡片、公众号排版等能力,流程和文章完全一样。文章链接不归这里管,用 article-fetch。
参数:$ARGUMENTS
1. 能力检查(本会话第一次)
node ${CLAUDE_SKILL_DIR}/scripts/doctor.mjs
看 video-input 一行:有没有 yt-dlp(YouTube、B 站)、ffmpeg(转音频)、本地语音识别(没有字幕时)。缺了的告诉用户怎么装,不要自己安装或下载。只有 .srt / .vtt 时什么都不需要。
2. 转写
node ${CLAUDE_SKILL_DIR}/scripts/transcribe.mjs "<视频链接或文件>" # → ./article2media-out/<平台>-<id>/
node ${CLAUDE_SKILL_DIR}/scripts/transcribe.mjs "<链接>" --prompt "人名 术语" --terms 词表.txt
node ${CLAUDE_SKILL_DIR}/scripts/transcribe.mjs ./talk.mp4 --title "标题" --author "作者"
- 文字来源:人工字幕优先,其次平台的原语言自动字幕(不用机器翻译的字幕轨),都没有时本地 Whisper 识别。已有的逐字稿默认复用,
--force重来。 - 专有名词:先用
--prompt "人名 术语"提示,再用--terms 词表.txt(每行错词=>正词)替换。 - 平台要求浏览器 cookie 时(YouTube 的机器人校验、抖音分享页不带数据):先问用户能不能复用本机浏览器状态,同意后才加
--cookies-from-browser chrome;不同意就请用户把视频保存到本地,直接传文件。 - 其他参数:
--lang识别语言、--keep-video保留 MP4(截图用)、--out指定目录。来源、清理、产出文件的细节见references/video-input.md。
3. 核对
- 完整读一遍
article.md,对照transcript/segments.md的上下文核对人名、作品名、术语。识别明显不对的,用--prompt/--terms重跑;拿不准的不要猜,交付时列出来。 - 逐字稿只补标点、分段,不改字。不要手工"润色"它——要整理成文章,是改写(
article-rewrite)的事。
4. 交付
单独使用时:article.md 和 transcript/segments.md 的绝对路径、平台、时长、文字来源(人工字幕 / 自动字幕 / 本地识别)、段数与字数,以及拿不准的专有名词。
在组合里:把目录交给下一个能力(见 ../article2media/references/pipeline.md)。发布由逐字稿做成的图文或视频时,来源写成「内容整理自 <作者> 的<平台>视频《标题》」。
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 464k | Updated | Claude CodeCodex | Verified | Install |
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 464k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 318k | Updated | Claude CodeCodex | Unverified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 318k | Updated | Claude CodeCodex | Unverified | Install |