The VAE audio quality is fantastic. Could you please open-source the weights and code of the VAE encoder in the future? I noticed that VAE training uses a multi-scale discriminator. Is it a multi-band STFT discriminator or Melgan's multi-scale discriminator? Could you provide the weights of each loss during training? Thank you so much!
The VAE audio quality is fantastic. Could you please open-source the weights and code of the VAE encoder in the future? I noticed that VAE training uses a multi-scale discriminator. Is it a multi-band STFT discriminator or Melgan's multi-scale discriminator? Could you provide the weights of each loss during training? Thank you so much!