The goal of this project was to identify 10 different genres of music using machine learning. The dataset used was GT-ZAN, and the genres were:
- Blues
- Classical
- Country
- Disco
- Hiphop
- Jazz
- Metal
- Pop
- Reggae
- Rock
The spectrogram of all the audio files were made and then were fed to the NN as input.
PyTorch was used to make the Neural Net, it had 3 convolutional layers, and four linear layers each with ReLu activation function.
Maximum test case accuracy obtained was 70.7%