Inspiration

I've always loved listening to music of various genres, but I do not possess the skills or knowledge to compose new music. This enables me to be a music creator of some form.

What it does

Mad Muzzic allows a user to submit midi files to the website, and a neural network will generate a custom midi file from their file inputs.

How I built it

The frontend of the website was created in React.js, Bulma, HTML5. The machine learning part was implemented using an LTSM neural network in Python using Music21 and Keras. The backend was done in Python and Flask using Google Cloud.

Challenges I ran into

We had a difficult time getting the neural network to train on AWS because of the resources required. We solved this by using Google Cloud. We are still working on integrating the frontend and the backend.

Accomplishments that I'm proud of

This is the biggest thing Eric has done in React.js. This is Anand's first time working with neural networks. Leonard is proud of writing API's in Flask and Google Cloud.

What I learned

We learned how to implement the front end and back end of a website to accomplish a task (in this case running a neural network). We also learned a lot about the workings of LTSM neural networks.

What's next for Mad Muzzic

Integrating back/front end mainly.

Built With

Share this project:

Updates