Re-implementation of the main convolutional model structure from Rachel Bittner's Basic Pitch model in PyTorch. Also Transfered pre-trained weights to PyTorch for integration in my Polyphonic DDSP implementation