Open skypro1111 opened 10 months ago
Adding the data with the correct dataset will work fine!
@pawel-polyai i think we need to change espeak backend language to support the phoneme sequence of the new languages. Also, how can we finetune it? as training from scratch gives Cuda oom.
Interested in training in other languages too. @pawel-polyai I'm interested in changing espeak to portuguese, since it supports. Do you think that by just changing the espeak language, and generating the dataset phoneme using espeak is a good way to start?
please tell me, is it enough to generate the correct dataset with the correct manifest for training in other languages, or is there some additional manipulation required?