The first hurdle we faced was finding how to accurately generate a fingerprint that could help us identify the song even if we were only using a small part of it with considerable noise.

First we used hashing but even a single bit chain can change the hash thus hashing was not appropriate for our usecase.

We finally found a finerprinting api that convert music to a fingerprint that can detect if its the same song even if it was trimmed and manipulated.

Next our project needed to use a blockchain so we decided to use IPFS but once we had fully integrated IPFS into our codebase with all of the base cases working, we realized that IPFS does not really have the concept of ownership and it instead focuses mainly on de duplication by generating a hash for each file uploaded and not allowing reuploading the exact same file again.

So we decided by storing our files with specific metadata we could solve this problem.

We also faced issue when we had to check the blockchain for our fingerprint that was stored and we had to iterate through each block thus increasing the complexity and resources used.

So we decided to use a local lookup table with content mapped