AI Billboard Hot 100

Built in collaboration with Nabil Baugher

One night on the fourth floor of my MIT dorm, I played some AI covers of popular songs for my friends. We ended up listening to AI covers for 3 hours straight and then decided to make a platform where people could discover and listen to the most popular AI music.

This project took place over the course of 1 week, and we got a bunch of users across MIT and had some of the top AI channels uploading music onto our platform.

Unfortunately, we ran into a bunch of issues with covers constantly being taken down from YouTube and shortly thereafter got an email from UMG asking us to take down the site.

Here were some of the technical highlights of the project:

Data Collection Pipeline

• YouTube API Integration:

  • Used the YouTube Data API v3 to search for channels with "AI cover" or "AI voice" in their titles or descriptions
  • Implemented a Python script that ran daily to fetch the latest videos from these channels
  • Created a scoring system based on views, likes, and comments to rank videos
  • Stored metadata in a MongoDB database for quick retrieval and updates

• TikTok API Integration:

  • Leveraged the unofficial TikTok API wrapper to access trending content
  • Built a Node.js service that monitored hashtags like #AIVoice, #AICover, and #AIMusic
  • Implemented rate limiting to stay within API constraints
  • Created a scoring algorithm that weighted recent engagement more heavily than older content

• Stream Count Tracking:

  • Developed a function that manually updated stream counts every 12 hours
  • Created a simple database to store metrics between updates
  • Implemented a notification system that alerted us when videos were taken down
  • Built a basic dashboard to visualize growth trends across different AI voices

User Experience

• Song Hosting:

  • Created an iframe-based player that embedded YouTube and TikTok videos directly
  • Implemented a fallback system that would try alternative sources if a video was removed
  • Added a custom player UI that maintained consistent styling across platforms
  • Built a simple caching mechanism to improve load times for frequently accessed content

• Interactive Charts:

  • Developed a simple charting system for displaying rankings
  • Created three main charts: Top Covers, Top Originals, and Viral
  • Implemented manual updates every 12 hours to refresh the charts
  • Added basic sorting options by platform

• Artist Pages:

  • Built individual profile pages for each AI voice model
  • Created a simple view showing the most popular covers for each voice
  • Implemented a basic recommendation system that suggested similar AI voices
  • Added simple analytics showing performance metrics