🎥 Ask Questions to Any YouTube Video Using RAG

This project demonstrates how to ask natural-language questions about the content of any YouTube video using Retrieval-Augmented Generation (RAG). We combine LangChain and Cohere embeddings to extract transcripts, perform semantic search, and generate answers based on the video.

🚀 Features

• ✅ Automatically extract subtitles or transcribe video audio using Whisper.
• ✅ Split and embed transcripts into vector representations.
• ✅ Retrieve the most relevant text chunk using semantic similarity.
• ✅ Generate an answer with a Large Language Model (LLM) based on the retrieved content.

📦 Requirements

• LangChain
• Cohere
• Chroma or another vector store