In this project I build a self-supervised binary emotion classifier from speech audio. Emotions are complex multidimensional concepts, but in this project, given an audio clip the model predicts whether the emotion of the voice is positive or negative.