Text this: Fusing Spatial and Temporal Features Extracted Using Convolutional Neural Networks and Gated Recurrent Units for Improved Deepfake Detection.