BeatsMe

Inspiration

Recent Advances in Convolutional Neural Networks to generate Speech and existing work from multiple groups

What it does

Based on a seed generates a set of beats that can be later put into a MIDI file

How I built it

Used Tensor Flow and Magenta to train a convolutional NN with a few hundred electronic drum beats

Challenges I ran into

Didn't have a UI so I have to use a template, it was hard to find a use case

Accomplishments that I'm proud of

A successful training of the network and the conversion to MIDI files

What I learned

Machine Learning is tough

What's next for BeatsMe

I already enrolled a couple of producers and artist willing to test it in real life scenarios.

Built With

• javascript
• tensorflow