video-to-text
Skill2 starsChanged 3 months agoMITView on GitHub
视频内容情报。用户发抖音分享链接(自动下载+语音转写全文)、B站链接、或本地视频/音频文件时使用。可输出文字稿、SRT/VTT 字幕(带真实时间戳),转写后可用 LLM 生成内容摘要和爆款结构拆解。中文转写默认输出简体+标点并按句切分。Video to text: Douyin link → auto download → transcribe → subtitle (SRT/VTT) → viral-content analysis in one command.
- Reads credentials
5 other projects use a version of this, 1 of them edited.
Adopted by
Most recently taken up by
- lgldlk/lgldlk-agent-skillstheir own version, first seen Oct 13, 2026
- jiawood2006/video-to-textedited, first seen Oct 13, 2026
- Luo4507/video-to-texttheir own version, first seen Sep 28, 2026
- Lotus-7/video-to-text-skilltheir own version, first seen May 8, 2026
- yfge/video-skills-suitetheir own version, first seen Feb 28, 2026
Other versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 1
- jiawood2006/video-to-text★ 019% same
Wrote their own version · 4
- lgldlk/lgldlk-agent-skills★ 3same name
- yfge/video-skills-suite★ 1same name
- Lotus-7/video-to-text-skill★ 0same name
- Luo4507/video-to-text★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository8 other files it gives its agents.
- ai-project-advisorSkill
- ai-video-ad-productionSkill
- de-ai-writerSkill
- doc-ocrSkill
- ecommerce-material-studioSkill
- memory-graphSkill
- memory-managerSkill
- voice-personaSkill
The file
Files
Video-to-Text — 抖音/本地视频 → 文字 → 内容情报(一条命令闭环)
抖音分享链接 → 自动下载 → 语音转写 → 结构化 Markdown → 爆款拆解,一条命令跑完。 不用手动存视频、不用第三方 API key、隐私不出本机(默认本地 faster-whisper)。
📁 安装:
hermes skills install jiawood2006/hermes-skills/skills/video-to-text或./install.sh(一键装+全局命令vtt)→ 脚本在~/.hermes/skills/utilities/video-to-text/scripts/。
触发条件
用户发送以下内容时使用本技能:
- 抖音分享链接(v.douyin.com / www.douyin.com/video/…),要"转文字/提取文案/拆解/分析爆款/看内容"
- B站链接,要"看视频信息"
- 本地视频文件(.mp4/.mov/.m4a),要"语音转文字/字幕/摘要/分析"
使用步骤
1. 抖音链接 → 全自动闭环(默认行为)
# 一条命令:元数据 + 自动下载 + 本地转写
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py "https://v.douyin.com/xxxx/"
# 全流程 + LLM 摘要 + 爆款结构拆解
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py "https://v.douyin.com/xxxx/" --all
# 只要元数据(不下不转)
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py "https://v.douyin.com/xxxx/" --meta-only
# 转写引擎换硅基 SenseVoice(中文口语/带噪更准,需 SILICONFLOW_API_KEY,失败自动回退本地)
SILICONFLOW_API_KEY=sk-xxx python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py "https://v.douyin.com/xxxx/" --engine sensevoice技术链路:curl 短链取真实视频 ID → Playwright 监听 aweme/detail API 拿元数据+无水印 play_url → 自动下载 → 提取音频 → 转写。
2. 本地视频文件 → 转写 + 内容情报
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py 视频.mp4 # 只转写
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py 视频.mp4 --summary # + 内容摘要
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py 视频.mp4 --analyze # + 爆款结构拆解
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py 视频.mp4 --engine sensevoice # 换 SenseVoice3. B站链接 → 元数据(零依赖公开 API)
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py "https://www.bilibili.com/video/BV1..."输出标题/作者/时长/播放/点赞/简介;转写请下载后走本地文件。
4. 输出产物(结构化 Markdown)
默认保存 <标题>_transcript.md,可直接进 Obsidian/知识库:
---
platform: douyin
url: https://v.douyin.com/xxx
title: 视频标题
author: 作者名
duration_s: 42
digg: 12345
date: 2026-09-04
tags: ["话题1", "话题2"]
---
# 视频标题
| 字段 | 值 |
|:---|:---|
| duration_s | 42 |
...
## 转写全文
[00:00] 大家好今天给大家…
[00:05] 这款产品…5. 输出字幕文件(SRT / VTT,带真实时间戳)
想剪视频、上传 YouTube/B站字幕、或做逐句校对时,直接出字幕文件:
# 文字稿 + 附带 SRT 字幕(最常用)
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py 视频.mp4 --subtitles srt
# 同时出 SRT 和 VTT
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py 视频.mp4 --subtitles srt,vtt
# 只要字幕,不要文字稿
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py 视频.mp4 --format srt- 输出
<标题>_transcript.srt(剪映/Premiere/YouTube 通用)或.vtt(网页播放器通用) - 中文默认输出简体 + 标点,并按句切分(不会一条字幕横跨十几秒)
- 切分阈值可用环境变量调:
VTT_MAX_CHARS=18(每条约多少字)、VTT_MAX_DUR=7(每条最长秒数)
6. 指定转写语言
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py 英文视频.mp4 --asr-lang en缺省 zh(普通话)。中文字幕会额外注入提示词引导简体输出。
7. 对已有转写文本做分析
python3 ~/.hermes/skills/utilities/video-to-text/scripts/analyze.py 转写.md --all--summary输出:一句话概括/核心要点/金句/目标受众/可借鉴点--analyze输出:开头钩子(0-10s)/节奏结构占比/情绪曲线/卖点/CTA/可复用套路--format txt输出纯文本(旧格式兼容);--out-dir指定目录
验证(安装后自测)
# 本地文件快速自测:没有视频?用 macOS 自带语音合成一段
say -v "Ting-Ting" "大家好,这是一段测试语音" -o /tmp/t.m4a # Linux 可跳过
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py /tmp/t.m4a --out-dir /tmp/vtt_test
# 期望:/tmp/vtt_test 出现 *_transcript.md 且含"测试语音"附近文本
# 字幕功能自测(应得到简体带标点、按句切分的 SRT)
python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py /tmp/t.m4a --subtitles srt --out-dir /tmp/vtt_test
# 期望:/tmp/vtt_test 出现 *_transcript.srt,含 --> 时间戳行依赖(首次使用时安装)
pip3 install faster-whisper # 默认本地转写(首次下载 tiny 模型 ~75MB)
# macOS 自带 avconvert;Linux 需 ffmpeg
# 抖音解析需 playwright + chromium(douyin_extract 内部使用)
# 内容摘要/爆款拆解需 LLM key(LLM_API_KEY 环境变量 或 ~/.deai_writer.conf)
# SenseVoice 引擎可选:SILICONFLOW_API_KEY(硅基流动,有免费额度)已知陷阱
- duration_ms=0 假阴性:SSR 元数据时长可能为 0 但实际是视频。有下载 URL 先下载验证,勿直接判图文。
- play_addr URL 过期:SSR 解析出的视频 URL 可能 404,此时下载失败会提示手动提供文件(自动降级不中断)。
- 图文内容:duration=0 且下载文件 <1MB 是图文(无音频),只输出元数据,不编造内容。
- whisper 转写慢:Intel Mac 3 分钟视频约 5 分钟;已自动
language='zh'+ VAD 过滤静音。要更快/更准用--engine sensevoice。 - 中文默认输出繁体(已修):whisper 对中文常输出繁体、且几乎不加标点。本技能已默认注入
initial_prompt引导简体 + 标点;如需自定义,设环境变量WHISPER_INITIAL_PROMPT(置空则关闭该引导)。 - 准确率取决于模型大小:默认
WHISPER_MODEL=tiny(~75MB,快但易错字)。要可用的转写质量就换small/medium:
首次会下载模型(国内网络建议WHISPER_MODEL=small python3 ~/.hermes/skills/utilities/video-to-text/scripts/vtt.py 视频.mp4 --subtitles srtexport HF_ENDPOINT=https://hf-mirror.com)。 - 字幕切分粒度可调:默认按标点/18 字/7 秒切条。要更碎或更整用
VTT_MAX_CHARS、VTT_MAX_DUR环境变量。 - SenseVoice 路径无真实时间戳:该引擎不返回时间信息,时长按语速估算,字幕时间轴仅供参考——要精确字幕请用 faster-whisper。
- 分析需 LLM key:--summary/--analyze 需 LLM_API_KEY;未配置时只转写不分析(明确降级不报错)。
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

