Can I combine multiple speaker weights into a single file?

[lextoumbourou]

I have a G_*.pth file with only one speaker id: "lex": 0.

I have a different G_*.pth with only one speaker id: "jonno": 0.

I want to combine them to have a single G_*.pth file with 2 speakers:

"lex": 0,
"jonno": 1

Is this possible?

Thanks for your help.

[hataori-p]

From what I understand about the model, I think it is not possible.
The voices should be trained jointly to make one common model.

There are speaker embeddings, which determine the speaker in the model, but I don't thing they are transferable between different models. See discussion here #282