💬 Talk-To-Tube: A YouTube Video Chatbot

Ever wished you could ask a long YouTube lecture a question or get a quick summary of a tutorial without scrubbing through the timeline? This project transforms YouTube videos into interactive conversational partners.

Simply provide a video URL and start asking questions. Get answers that are sourced directly from the video's content, ensuring accuracy and relevance.

---

🚀 The Mission

This application uses a Retrieval-Augmented Generation (RAG) architecture to let you chat with the content of any YouTube video. It was built as a hands-on exploration to master the core components of modern AI systems: from ingesting documents and creating vector embeddings to performing similarity searches and prompting LLMs with grounded, contextual information. The goal was to build a system that relies on a specific knowledge base (the video) rather than just its pre-trained data.

---

🛠️ The Toolkit: Under the Hood

This project is built with a modern, efficient tech stack:

• 🧠 Core Logic:

  • Architecture: Retrieval-Augmented Generation (RAG)
  • LLM & Embeddings: Google Gemini API & Google Embeddings

• ⚙️ Backend Engine:

  • Framework: FastAPI
  • Server: Uvicorn

• 🖥️ User Interface:

  • Framework: Streamlit

• 📚 Knowledge Base:

  • Vector Store: FAISS (Facebook AI Similarity Search)

---

✨ Core Capabilities

• ✅ Seamless Content Ingestion: Just drop a YouTube URL to get started.
• ✅ Conversational Q&A: Ask questions in natural language.
• ✅ Grounded Answers: Responses are based only on the video's content, preventing off-topic answers.
• ✅ On-the-Fly Processing: Process videos and start chatting almost instantly.

---

⚡ Launch Sequence: Get Running in Minutes

❗️ Important: This project is designed for Python versions 3.8 to 3.13. We recommend using python3.13 for best compatibility.

1. 📥 clone the repo

git clone https://github.com/Bhr-007/Talk-to-tube.git

2. 📁 Navigate to project directory

cd Talk-to-tube

3. 🔑 Setup API Key

• Generate an API Key 🔗 Here.
• Add it to your .env file:

echo 'GOOGLE_API_KEY="[YOUR-API-KEY-HERE]"' >> .env

4. 📦 Install dependencies

pip install -r requirements.txt

5. 🔧 Start the backend server

uvicorn backend.main:app --reload

6. 🌐 Launch the frontend

streamlit run frontend/streamlit_app.py

---