I want to train the vocoder on my own data set. Is there any suggestion on the amount of data to converge, as well as the training time, and the loss at convergence.
ParallelWaveGAN copied to clipboard
kan-bayashi
