What makes models compatible?

I created my model with tacotron2 and pwgan using the mozilla repository.
Now, there are many repositories on github, for instance this very interesting one.
How do I figure out if my model is compatible with a random tacotron2 + pwgan fork before testing it out?
Do I just look for tacotron2 or do I need to know more?
I am asking because every fork seems to require you to start from zero, the only commonality seems to be the Ljspeech-dataset-like foundation.

I don’t understand what you mean. Do you like to use a mode trained on a repo in another repo?

It is normal that different repos have different conventions and thus are more compatible.

1 Like

I like to use a model trained on one repository in another repository.

there is no such standard that enables compatibility. It is not unfortunately possible.

1 Like

It is how Eren describes it. Usually there are also different sound processing attributes to begin with.