Classifies song clip as classical or rock using CNN
This project classified songs as either rock or classical based purely on 30 second audio clips. To achieve this, each audio clip was converted into an image called a spectrogram. The below graph is how audio clips are normally represented.
The following image shows the graph above converted into a spectrogram, which is a gray-scale image.
A convolutional neural network was trained on these images and achieved 71 accuracy on the test audio.
