This is a 288x288 wav2lip model version. Some Features I will implement here

• input size 288x288
• PRelu
• LeakyRelu
• Gradient penalty
• Wasserstein Loss
• Kubeflow for training process
• Orthogonal regularization with gradient penalty

I trained my own model on AVSPEECH dataset and then transfer learning with my private dataset.