FFTNet Pytorch implementation of Princeton & Adobe's FFTNet model from FFTNET: A REAL-TIME SPEAKER-DEPENDENT NEURAL VOCODER Disclaimer I am not an author of the paper.