✂️ FastCut — Video Transcript Editor

Upload a video → get a word-level transcript with millisecond timestamps → click words to mark them for deletion → export a cleanly cut video.

Auto-detects Chinese filler words: 嗯 额 啊 呃 哦 那个 就是 然后 etc.

---

Prerequisites

• Python 3.10+
• Node.js 18+
• ffmpeg (must be in PATH)
• An OpenAI API key

---

Setup

1. Backend

cd backend
pip install -r requirements.txt
export OPENAI_API_KEY=sk-...       # Mac/Linux
# set OPENAI_API_KEY=sk-...        # Windows
uvicorn main:app --reload --port 8000

2. Frontend

cd frontend
npm install
npm run dev

Open http://localhost:3000

---

How to Use

1. Upload your video (MP4, MOV, AVI, MKV)
2. Click Transcribe with Whisper — Whisper returns every word with ms-level timestamps
3. In the editor:
   • 🟡 Orange chips = auto-detected filler words (嗯 额 etc.)
   • Click Auto-mark all fillers to mark them all at once
   • Click any word chip to toggle it for deletion (turns red with strikethrough)
   • Use the Search & delete box to find and cut a specific word everywhere
   • Click any word to also jump the video to that moment
4. Click Export → ffmpeg cuts the marked segments → download the clean video

---

Filler Words Auto-Detected

Chinese: 嗯 额 啊 呃 哦 噢 唔 哎 哼 那个 就是 然后 这个 嘛 吧 呢
English: uh um erm hmm ah oh

Add more in backend/main.py → FILLER_WORDS set, and frontend/src/TranscriptEditor.jsx → FILLER_WORDS set.

---

Project Structure

video-transcript-editor/
├── backend/
│   ├── main.py           # FastAPI server
│   ├── requirements.txt
│   ├── uploads/          # auto-created: incoming videos + transcripts
│   └── outputs/          # auto-created: exported videos
└── frontend/
    ├── src/
    │   ├── App.jsx           # Upload, loading, done screens
    │   ├── TranscriptEditor.jsx  # Core word editing UI
    │   └── App.css           # Dark editorial design
    ├── index.html
    ├── package.json
    └── vite.config.js