Minimum data requirements and inference without training.

[Tortoise17]

Dear Friends, I have questions:

1. I want to know how much minimum data is required for any new speaker training? and is it possible to train that speaker for some other language?
2. Also if the new language training needs completely new model customization, how much data is required to train such model?
3. Is it possible or any way to make inference of any new speaker using the profile sample and generating the emotion for that without prior training?

Please guide and help.