This project implements a Retrieval-Augmented Generation (RAG) system that accepts audio or video input, transcribes it, and performs question-answering using Google Gemini and Qdrant.
- RAG (Retrieval-Augmented Generation) pipeline using LangChain and Gemini
- Support for both audio and video input
- Semantic search backed by Qdrant vector database
- Dual response generation:
- 🔍 Search-based context
- 🤖 LLM-powered Gemini response
- Transcription of
.mp3or.wavfiles using Faster-Whisper - High-quality offline transcription with fast performance
- Automatic audio extraction from videos using FFmpeg
- Full transcription of video content for downstream processing
- Semantic embedding using Google Generative AI (Gemini embeddings)
- Vector storage and similarity search using Qdrant
- Can run entirely locally (without OpenAI or cloud dependencies)
- Ideal for constrained environments with privacy concerns
- Built in Jupyter Notebook for transparency, experimentation, and research workflows
| Technology | Logo | Purpose |
|---|---|---|
| Python | Backend logic and orchestration | |
| LangChain | RAG architecture and LLM interface | |
| Gemini (Google Generative AI) | LLM and Embeddings | |
| Qdrant | Vector database for semantic search | |
| Faster-Whisper | Fast local speech-to-text model | |
| FFmpeg | Audio extraction from video files | |
| Jupyter Notebook | Interactive development environment |
- Python = 3.10
- FFmpeg (added to PATH)
- Gemini API Key
- Install dependencies:
- pip install -r requirements.txt
-
Download FFmpeg:
- Visit: https://www.gyan.dev/ffmpeg/builds/
- Under “Release builds”, download
ffmpeg-release-full.7zorffmpeg-release-essentials.zip
-
Extract the Archive:
- Use 7-Zip or WinRAR to extract the file
- Extract the contents to
C:\ffmpeg
-
Add to System PATH:
- Open:
Control Panel→System→Advanced system settings→Environment Variables
- Under System variables, find
Path→ Click Edit - Click New and enter:
C:\ffmpeg\bin - Click OK to save
- Open:
-
Verify Installation:
- Open Command Prompt and run:
ffmpeg -version
- Open Command Prompt and run: