Purrsona allows creating personalized Spotify playlists for pets. For this, the user uploads a picture of their pet and Purrsona analyzes the breed of the pet and creates a personalized playlist based on the song history of the user and breed information.

The project has mainly been created as a fun side project to learn more about the Spotify API and machine learning. The underlying algorithm for creating playlist checks a list of attributes contributed to the breed of the pet and creates a playlist of songs that match these attributes.

Lessons learned

A large challenge in creating Purrsona was creating a machine learning model that has a possibility of being accurate while still being small enough to be run in the browser.

As a goal, the model didn't need to be accurate in all cases, but instead having a chance of getting a correct or at least similar result in 50% of cases as it would give users a sense of the algorithm working and actually being tailored to their pet.

Build with

Build with NextJS, Teachable Machine and the Spotify API.