YouTube Notes Generator with Gemini & ChromaDB

This project automatically extracts transcripts from YouTube videos, summarizes them using Google's Gemini 1.5 Flash model, and stores the results in a persistent Chroma vector database for later retrieval and querying.

Features

• Download and parse YouTube video transcripts
• Summarize transcripts using Gemini 1.5 Flash
• Store notes in a Chroma vector database with vector embeddings
• Easily search or expand later using vector search

Technologies Used

• Google Generative AI
• ChromaDB
• youtube-transcript-api
• Python 3, Google Colab
• SQLite3, .bin index files
• Google Drive integration

How It Works

1. The user provides a YouTube video ID.
2. The transcript is fetched using youtube-transcript-api.
3. Gemini generates a summary based on the transcript and a customizable prompt.
4. The result is saved as .txt and stored in a Chroma vector collection with an embedding function.

Folder Structure

yt-notes/
├── get_video_notes.ipynb  # Main Colab notebook
├── chroma.sqlite3         # Metadata for vector DB
├── index/                 # Binary index files (auto-generated)
├── temp_notes.txt         # Summarized notes (temporary)
└── temp_transcript.txt    # Transcript from YouTube video (temporary)

Environment Variables

Secrets must be added securely. For Colab, use the Secrets tab or userdata.get():

• GEMINI_API_KEY - your Google AI API key
• CHROMA_GOOGLE_GENAI_API_KEY - used for embedding functions

from google.colab import userdata
GEMINI_API_KEY = userdata.get("GEMINI_API_KEY")

Setup (Colab)

1. Mount your Google Drive
2. Install dependencies (via !pip install)
3. Add your Secrets using the Colab sidebar
4. Run the notebook get_video_notes.ipynb

Contributions

PRs and ideas are welcome! Just open an issue or fork the repo.

Built by Hadeel❤️