Question: Using a pretrained encoder for getting the speaker embedding. about speechsplit

[terbed]

I have the same question @auspicious3000
Here you use the one-hot encoded embedding with a lent of 82 (the number of speakers it was pretrained), but could you generate a zeros-shot general embedding like in AutoVC. If I am correct the size of the used embedding was larger in that, I assume you cannot use that here.

So to wrap up: this method with the pretrained weights works only on the 82 speakers it was trained and conditioned on if we consider only the timbre conversion?

[FurkanGozukara]

@nischal-sanil did you make it work?

can you check my question please? #28

[auspicious3000]

@terbed Yes. Unless you retrain the model.