You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Why do you mel_loss use fmax=none for Melp instead of using input and generation?
The input (mel-spec) of the generator uses fmax=8000, while the mel-spec used in loss uses fmax=none, that is, sr/2.
What are the benefits of this?
The text was updated successfully, but these errors were encountered:
Why do you mel_loss use fmax=none for Melp instead of using input and generation?
The input (mel-spec) of the generator uses fmax=8000, while the mel-spec used in loss uses fmax=none, that is, sr/2.
What are the benefits of this?
The text was updated successfully, but these errors were encountered: