Do you think the results would be better if i saved the .npy spectrogram and used my own vocoder to inference it?
Do you think the results would be better if i saved the .npy spectrogram and used my own vocoder to inference it?