Blog Post 2

We are now half way into week three, our team has submitted the project plan and we have a (mostly) solid understanding of what needs to be done.

Our project is a music genre classification neural network. This project, at its core, is a Convolution Neural Network(CNN) that will be trained to classify music genres from a large data set of audio clips. For implementing this neural network we decided to use Keras for implementing our neural network because it seemed to be more beginner friendly compared to Pytorch. The data we will feed into our CNN will likely take the form of spectrogram images, to obtain these we will be using the popular audio processing library Librosa. We also hope to include a frontend for the end user to upload music clips to, if we reach this stretch goal it will likely be implemented using React.

So far most our work has been focused on researching neural networks, audio processing and cloud hosting. Go forward, some parts of this projects, such as setting up AWS for cloud hosting, UI, will be worked on independently. Other parts, such as implementing and training the CNN, aka the meat of the project, will require collaboration and input from all members.