Inspiration
As budding musicians, we've had our fair share of struggles with obtaining sheet music for pieces that we wanted to learn how to play. Sometimes the sheet music was hidden behind a paywall and sometimes, the sheet music didn't exist online. Our goal with OsKey is to increase accessibility to sheet music to allow musicians to work on their craft without worrying about obtaining the sheet music.
What it does
OsKey takes in audio recordings (MP3s), breaks it down to the individual notes, and uses machine learning to identify each distinct note. Then OsKey formats this into a list which is pushed to a database. Within that database, OsKey takes the notes and turn them into a beautiful visualization of the sheet music.
How we built it
We trained OsKey on a database of music notes and songs to boost its accuracy in recognizing the notes. We then converted the notes it found into lists which we pushed to a database. Within this database, we took these notes and converted them into visualizations of how the notes should look on sheet music.
Challenges we ran into
Some of the challenges included accessing and cleaning the dataset, lack of documentation for CreateML files, managing the dataset, and visualizing the data. Apart from this, we learned a lot through this challenge and had tons of fun.
Accomplishments that we're proud of
We were able to rapidly gain new technical ground in Firebase, CreateML, ReactJS, and more. Along with this, we were retrospectively proud of persevering throughout the night (a little caffeine goes a long way) and being able to work as a coherent unit to finish this project.
What we learned
We learned that there are always multiple approaches to the same problem, that we can have fun no matter what, and that we should ALWAYS set up our technical environment before we start a hackathon again.
What's next for OsKey
Refine the classification model, expand into more instruments (trumpet, alto sax, etc), and be able to offer a versatile tool to both burgeoning and advanced musicians.


Log in or sign up for Devpost to join the conversation.