E2E-V2SResNet: Deep residual convolutional neural networks for end-to-end video driven speech synthesis | Publicación