Using Fastspeech2 instead of Tacotron2 with LPCNet

Hi,

Thanks for this repo which helps to setup Tacotron2+LPCNet.
I am trying to setup Fastspeech2 with LPCNet based on this LPCTron repo. I am able to get audio synthesized but it seems to get clipped when the utterance is longer. Anyone can point out the issue here? or if I am doing something wrong with Fastspeech2+lpcnet combination?
-Sagar