balewgize / skimmit

Article and YouTube video summary from URL

Skimmit

Article and YouTube video summary from URL, powered by GPT-4 & Gemini Pro
View Demo

Skimmit is now available on GPT Store

About The Project

The main goal of this project is to make content consumption effective. How?

• Quickly get the main idea of any article or video.
• See a short preview to asses if it's worth your time.
• Spend less time consuming, and more time doing.

Why Skimmit?

I believe content is being consumed more than ever. So it's important to be selective and make the most of it.

If you want to give a try, check out the web version at https://balewgize.app/

Built With

• Python
• Flask
• LLM (GPT-4 & Gemini Pro)

Getting Started

To get a local copy up and running follow these example steps.

Prerequisites

• Python installed
• Gemini Pro API key

Installation

1. Clone the repo and navigate to skimmit directory

   git clone https://github.com/balewgize/skimmit.git
   

2. Install required packages (virtual environments recommended)

   python3 -m venv venv && source venv/bin/activate
   

   pip install -r requirements.txt
   

3. Provide Gemini API key in .env file

   GOOGLE_API_KEY=your-google-gemini-api-key
   

4. Start the server.

   flask run
   

5. Goto http://127.0.0.1:5000 on your browser

Thanks for checking!

skimmit's Issues

Use Whisper or Gemini to transcribe YouTube video

Cool project!

In my experiments, I found that Whisper (OpenAI speech-to-text model) produces vastly better results when transcribing videos, compared to the default YouTube subtitles. Gemini is supposed to be even better, although I haven't tried.

Would it be possible to support transcribing using a model, in addition to the default YouTube subtitles?