Argument x in WaveRNN's forward(self, x, mel) function
QdFei opened this issue · comments
Hi there,
I'm very new to WaveRNN, and trying to read the code. I found myself a little lost for the forward(self, x, mel) function in the WaveRNN class. I read a little bit more about the code and found that x seems to be a segment of the target compressed wav during training (and a vector of 0s during test). But I am not sure why it is used as the input of the model (concatenated with the input mel). Could anyone help to clarify the purpose of using this 'x' as input of the model? Thanks a lot.
already figured out. thanks