In this video, we will learn that we can use Principal Component Analysis (PCA) to find a transformation matrix that minimises reconstruction error for dimensionality reduction. We also show how this can be used to process speech clips for downstream machine learning tasks.