ajaiswal1008

I am training the voice conversion model from scratch using the Obama audio file. I have trained for around 20K steps and the loss is not decreasing much (recon0:0.03, recon:0.03,...