arthasjax / audioclassification

Audio MNIST Classification using 1D-CNN, 2D-CNN, GAN+2D-CNN, CVN+RandomForest, and LSTMs.

Audio MNIST Classification

Goal of Study: Classify Audio MNIST data. Please note that, audio data with label 2 has been removed from the training data.

Methods used: 1D-Convolutional Neural Network, 2D-Convolutional Neural Network, Convolutional Variational Autoencoder with Random Forest, Generative Adversarial Network with 2D-Convolutional Neural Network, and Long Short-Term Memory Network.

If you are using Google Colab: Please create a folder named "AudioClassification" in your GoogleDrive and save the data in that folder.

If you are running it on your computer please upload the data to Notebook manually.

You can download the data from https://www.kaggle.com/c/audio-classification/data

GAN with 2D-CNN model achieved the best performance while classifying test data.