Speak with PDF

Chat with multiple PDFs using AI-powered embeddings and conversational memory.

🚀 Vision

A smart AI assistant for students and professionals to extract insights from college textbooks, exam papers, and research documents, providing accurate answers with an interactive chat interface.

📌 Features

• Upload multiple PDF files and extract text
• Chunk text for efficient retrieval
• Use FAISS for vector-based search
• Leverage Google's Gemini AI for intelligent responses
• Maintain conversation history for better context
• Optimized for educational and research purposes

🛠️ Tech Stack

• Python (Core logic)
• Streamlit (Frontend UI)
• FAISS (Vector database for fast search)
• Google Generative AI (LLM and embeddings)
• LangChain (Efficient AI pipeline management)
• PyPDF2 (PDF text extraction)

📦 Installation

# Clone the repository
git clone https://github.com/your-username/speak-with-pdf.git
cd speak-with-pdf

# Create a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows use `venv\Scripts\activate`

# Install dependencies
pip install -r requirements.txt

🔑 Environment Variables

Create a .env file and add your Google API key:

GOOGLE_API_KEY=your_google_api_key

🚀 Running the Application

streamlit run app.py

🖼️ Usage Guide

1. Upload one or more PDF files from the sidebar.
2. Click on the Process button to extract and analyze text.
3. Ask questions about the content and receive intelligent responses.

🤖 How It Works

1. Extract PDF text using PyPDF2.
2. Chunk text into manageable pieces for better search performance.
3. Create embeddings using Google's Generative AI.
4. Store vectors in FAISS for quick retrieval.
5. Enable conversation with memory using Gemini AI.

📌 Future Enhancements

• Add support for other document types (DOCX, TXT, etc.).
• Improve UI with better visualization.
• Deploy on cloud platforms (Streamlit Sharing, AWS, or GCP).
• Integrate speech-to-text for voice-based interaction.
• Enhance AI model for better context understanding.

📜 License

MIT License

🤝 Contributing

Pull requests are welcome! If you have suggestions or improvements, feel free to fork the repo and submit a PR.

🌟 Acknowledgments

• LangChain for efficient LLM-based workflows.
• Streamlit for an easy-to-use UI.
• Google Generative AI for embeddings and chat capabilities.