Thanks for nice work. @bshall

I'm trying to train hifigan now, but it takes so long training it from scratch using other dataset.

If discriminator of base model is also available, I could start finetuning based on that vocoder. it seems that you released only generator. Could you also release discriminator weights?