You see a great Douyin video. You want to repurpose its content for Xiaohongshu, or turn it into a narrated video of your own. Normally you'd: watch, pause, transcribe by hand, rewrite, record voiceover, find stock footage, edit... hours of work.
WHALE automates the entire pipeline in minutes.
| Manual Work | WHALE Automation |
|---|---|
| Watch, pause, hand-copy script | 🎬 Puppeteer extracts in 10s |
| Rewrite, brainstorm hooks, titles | ✍️ DeepSeek dual-style rewrite |
| Find voice talent, re-record | 🎙️ edge-tts with 6 Chinese voices |
| Source footage, edit, subtitle | 🎥 Pexels + ffmpeg auto-synthesis |
| # | Stage | Tech | Time | Output |
|---|---|---|---|---|
| 1 | 🎬 Extract | Puppeteer | ~10s | Title, author, engagement data |
| 2 | 📝 Transcript | faster-whisper | 2–5min | Full speech-to-text |
| 3 | ✍️ Rewrite | DeepSeek | ~5s | Xiaohongshu / Douyin style |
| 4 | 🎙️ TTS | edge-tts + SSML | ~10s | MP3 audio |
| 5 | 🎥 Video | Pexels + ffmpeg | ~30s | 1080×1920 vertical MP4 |
| 💄 Xiaohongshu | 🎵 Douyin Select | |
|---|---|---|
| Title optimization | ✅ SEO keywords | ✅ 3-second hook |
| Body structure | 7 templates | TTS-friendly |
| Anti-AI-tone | ✅ 18 hook types | ✅ Conversational |
| Emoji | ✅ | ❌ (TTS reads them aloud) |
| Ending | Engagement ask | Follow + bookmark |
git clone https://github.com/Rookage/whale.git
cd whale
npm install
pip install edge-tts faster-whisper
cp .env.example .env # Add your DeepSeek API Key
npm start # http://localhost:3000
# Optional: add Pexels API Key in .env for video images| Layer | Tech |
|---|---|
| Browser Extraction | Puppeteer v25 |
| Speech Recognition | faster-whisper small |
| AI Rewrite | DeepSeek API |
| Text-to-Speech | edge-tts v7.2.8 + SSML |
| Image Search | Pexels API |
| Video Synthesis | ffmpeg v8.1.1 |
| Frontend | Vanilla HTML/CSS/JS, Mario NES pixel-art theme |
MIT
做自媒体的都知道：看别人视频很简单，把自己的想法变成视频很痛苦。 WHALE 把最花时间的文案提取、改写、配音、配画面全部自动化。
| 手工操作 | WHALE 自动化 |
|---|---|
| 看视频、暂停、手抄文案 | 🎬 Puppeteer 10 秒提取 |
| 自己改写、想标题、找钩子 | ✍️ DeepSeek 双风格改写 |
| 找配音、调语速、反复录 | 🎙️ edge-tts 6 种语音 |
| 找素材、剪辑、加字幕 | 🎥 Pexels + ffmpeg 自动合成 |
| # | 阶段 | 技术 | 耗时 |
|---|---|---|---|
| 1 | 🎬 视频描述提取 | Puppeteer | ~10s |
| 2 | 📝 逐字稿生成 | faster-whisper | 2-5min |
| 3 | ✍️ AI 改写 | DeepSeek（小红书+抖音双风格） | ~5s |
| 4 | 🎙️ TTS 语音 | edge-tts（6 种中文语音） | ~10s |
| 5 | 🎥 视频合成 | Pexels + ffmpeg | ~30s |
| 💄 小红书 | 🎵 抖音精选 | |
|---|---|---|
| 标题优化 | ✅ SEO 关键词 | ✅ 3 秒钩子 |
| 正文结构 | 7 种模板 | TTS 友好 |
| 去 AI 味 | ✅ 18 种钩子 | ✅ 口语化 |
| emoji | ✅ | ❌（TTS 会读出来） |
4 个马里奥角色精灵 + 绿色管道流水线 + 金币黄焦点高亮。零框架，原生 HTML/CSS/JS。
git clone https://github.com/Rookage/whale.git
cd whale
npm install
pip install edge-tts faster-whisper
cp .env.example .env # 填入 DeepSeek API Key
npm start # http://localhost:3000| 层 | 技术 |
|---|---|
| 浏览器提取 | Puppeteer v25 |
| 语音识别 | faster-whisper small |
| AI 改写 | DeepSeek API |
| 语音合成 | edge-tts v7.2.8 + SSML |
| 视频合成 | ffmpeg v8.1.1 |
MIT
Built with ❤️ by Rookage