CogniTube

Unleash the power of AI on YouTube!

Comment

Inspiration

Over the past few months, AI has been booming and new project are coming out day by day. ChatGPT, Nvidia’s Magic3D, Stable diffusion and more awe-some projects are released. I just came an idea with why not implement AI in videos. The goal is to create a complete AI enhanced platform for videos.

What it does

CogniTube is an application that implements AI to the video and provides transcript, summary, resources, translation, AI generated blog and export options.

How I built it

Initially I've created a GitHub project board to keep track on what to do. Then I've implemented each feature one-by-one with the following technologies:

  • youtube-dl - To download YouTube videos
  • Deepgram - Implemented Automatic Speech Recognition (ASR) technology to extract transcript, summary and topics.
  • Streamlit - Time's running! So for faster UI developement, Streamlit comes into play.
  • OpenAI - With the next-gen powerful AI models, we are able to create a blog posts based on the transcript and also extract the relevant information mentioned in the video.
  • Gitpod - It provides me a complete dev environment.

Challenges we ran into

  • Parsing JSON data - Thanks to StackOverflow!
  • Exposed the API Key - GitGuardian informed it!

Accomplishments that we're proud of

  • TL;DW - Too Long Did'nt Watch | Now you can read what's the video is all about!
  • Translate to different languages | வெவ்வேறு மொழிகளில் மொழிபெயர்க்க முடியும்
  • Topics, Keywords, and external links - Extracts brands, topics and keywords
  • Create AI generated blogs - Blogs reimagined!
  • Export to audio and PDF - Share as you wish!

What I've learned

  • Learned about ASR
  • Explored API's
  • Some new Python techniques and concepts
  • Streamlit features

What's next for CogniTube

The translation can be done using the AI models like Facebooks's fairseq MMT model or Hugging face models. Also we can implement AI to create presenation slides from videos. IDK, but maybe we can create a speech translation device powered with TinyML.

  • Platform for content creator to work on their videos
  • Use AI to create content and simplifies their work
  • Integrate various platforms
  • Analyze feedbacks and audiences

The thumbnail is created by AI

Built With

Share this project:

Updates

Submission history