Objective function for fast speech model #102
Description
Hi,
I did not see information about the objective loss for the FastSpeech model (the duration prediction's loss is indeed the MSE loss). Where did you get this information? Is the loss for the FastSpeech MSE loss?