Input song by the wonderful Mr.305!
Recommendations served by the system
Being a hobby musician myself(Michael), I noticed that the traditional streaming services such as Spotify, Apple Music, and SoundCloud do not cater to the average artist, who usually has little to no fanbase and is therefore not included in traditional recommendation systems.
What it does
Sonify relies on artificial intelligence techniques to analyze music and recommend music based on similarity and user feedback, versus a solely collaborative filtering based system. It uses the InterPlanetary File System (IPFS) to host the actual music, and that hash is stored in the ethereum blockchain using smart contracts. Recording stream counts and statistics on smart contracts would allow artist payouts without a middleman, and IPFS allows for a much cheaper storage solution and higher payouts for artists, a huge problem in the industry.
How we built it
This started off as a school project, solely as a music recommendation system. We used deep learning libraries such as tensorflow to build models for music analysis. We ended up winning first place for our project at the University of Miami and our professor made us aware of this competition. We then utilized the plethora of ethereum and web3 libraries to implement smart contracts written in solidity.
Challenges we ran into
Putting a project of this caliber live requires a massive amount of infrastructure setup, which the provided timeframe did not allow us to complete. We plan on continuing this further even after the competition ends. Currently the project is locally based, and so we can only put pictures of the demo.
Accomplishments that we are proud of
The incredible amount of skills developed to produce this demo.
What I learned
Developed skills in blockchain development, as well as learning a new language.