.github/copilot-instructions.md

AudioDubb Project - Development Instructions

Project Overview

AudioDubb is an AI-powered multilingual audio dubbing engine optimized for Hugging Face Spaces deployment. It transcribes, translates, and generates dubbed audio while preserving speaker identity and emotional expression.

Checklist

• Verify copilot-instructions.md file created
• Clarify Project Requirements (Hugging Face Spaces audio dubbing app)
• Scaffold the Project
• Customize the Project
• Install Required Extensions (Gradio built-in, no VS Code extensions needed)
• Compile/Verify the Project
• Create and Run Task
• Launch the Project
• Ensure Documentation is Complete

Project Structure Created

AudioDubb/
├── app.py                      # Main Gradio application
├── requirements.txt            # Python dependencies
├── .gitignore                  # Git ignore rules
├── README.md                   # Complete documentation
├── README_HF.md                # Hugging Face Spaces metadata
├── .github/
│   └── copilot-instructions.md # Development instructions
└── src/
    ├── __init__.py
    ├── models/
    │   ├── __init__.py
    │   └── model_manager.py    # Model loading and caching
    └── core/
        ├── __init__.py
        ├── transcriber.py      # Whisper Large v3 integration
        ├── translator.py       # NLLB-200 Distilled integration
        ├── voice_cloner.py     # XTTS-v2 voice cloning
        ├── audio_processor.py  # Audio I/O and processing
        └── pipeline.py         # Complete dubbing workflow

Implementation Details

Components Implemented

1. Model Manager: Singleton for efficient model caching with GPU support
2. Transcriber: Whisper Large v3 with language detection
3. Translator: NLLB-200 Distilled with batch processing
4. Voice Cloner: XTTS-v2 with speaker identity preservation
5. Audio Processor: File I/O, normalization, and cleanup
6. Dubbing Pipeline: Orchestrates complete workflow
7. Web Interface: Gradio app with advanced options

Key Features

• ✅ In-memory only processing (no permanent storage)
• ✅ GPU acceleration and model caching
• ✅ 20+ language support
• ✅ Speaker identity and emotion preservation
• ✅ Privacy-first architecture (no logging)
• ✅ Educational/personal use disclaimer
• ✅ Hugging Face Spaces optimized

Key Requirements

• ✅ Hugging Face Spaces ONLY deployment (no local storage)
• ✅ In-memory only processing (no permanent storage)
• ✅ Models: Whisper Large v3, NLLB-200 Distilled, XTTS-v2
• ✅ GPU acceleration and model caching
• ✅ Gradio web interface for HF Spaces
• ✅ Support 20+ languages
• ✅ Privacy: No logging, immediate cleanup
• ✅ Educational/personal use disclaimer
• ✅ Complete documentation for HF Spaces deployment

Files Created

• app.py - Gradio web interface (HF Spaces only)
• requirements.txt - Python dependencies
• README.md - Complete documentation
• DEPLOYMENT.md - HF Spaces deployment guide
• README_HF.md - HF Spaces metadata
• .gitignore - Git ignore rules
• src/models/model_manager.py - Model caching and management
• src/core/transcriber.py - Whisper transcription
• src/core/translator.py - NLLB-200 translation
• src/core/voice_cloner.py - XTTS-v2 voice cloning
• src/core/audio_processor.py - Audio I/O and processing
• src/core/pipeline.py - Dubbing workflow orchestration

Deployment Instructions

1. Create new Space on huggingface.co/spaces
2. Select Gradio SDK
3. Upload all project files
4. Application auto-deploys with HF infrastructure
5. See DEPLOYMENT.md for detailed steps