Hi, your code really helps a lot, I want to know that how can we predict the generataed Speech Command dataset results by trained model.