Logo Lanfrica

AbdoIR/Darija-Dataset-Builder

Domain:

natural language processing

Record type:

software
Creator:
Abd
Host:
A modern web application for creating labeled Darija audio datasets. # Darija Dataset Labeling Tool A modern full-stack application for creating labeled audio datasets with multiple input methods. The app supports YouTube videos, file uploads, and direct recording, providing a sleek web interface with real-time WebGL effects for labeling audio clips. ## Features - **YouTube Processing** - Download and split videos automatically - **File Upload** - Upload audio/video files for processing - **Direct Recording** - Record audio directly in the browser - **Modern UI** - Glassmorphism design with WebGL particle effects - **History View** - Track all labeled content by dialect category - **Progress Tracking** - See how many clips you've labeled - **Smart Deduplication** - Automatically skip duplicate entries - **Category Organization** - Organize by dialect (Urban, Northern, Saharan, Amazigh) ## Project Structure ``` darija-dataset/ ├── backend/ │ ├── app.py # Flask API server │ ├── requirements.txt # Python dependencies │ ├── dataset.csv # Generated labeled dataset │ ├── videos/ # Temp: Downloaded files (auto-deleted) │ └── clips/ # Audio clips organized by ID │ └── recordings/ # Direct recordings ├── frontend/ │ ├── src/ │ │ ├── App.jsx # Main application │ │ ├── App.css # Global styles with utility classes │ │ ├── components/ # React components │ │ │ ├── ViewToggle.jsx │ │ │ ├── InputSection.jsx │ │ │ ├── ModeSelector.jsx │ │ │ ├── CategorySelector.jsx │ │ │ ├── YouTubeInput.jsx │ │ │ ├── FileUpload.jsx │ │ │ ├── RecordingControls.jsx │ │ │ ├── LabelingSection.jsx │ │ │ ├── HistoryView.jsx │ │ │ └── PixelBlast.jsx # WebGL background effect │ │ └── lib/ │ │ └── utils.js # Utility functions │ ├── tailwind.config.js # Tailwind CSS configuration │ ├── postcss.config.js # PostCSS configuration │ └── package.json …