Low memory inference fix (#536)
* For low_mem, use spawned workers instead of forked workers (resolves #36) Used implementation from @lilydjwg: https://github.com/CorentinJ/Real-Time-Voice-Cloning/issues/36#issuecomment-529380190 * Different method of passing the seed for low_mem inference Resolves #491, #529, #535
Showing
想要评论请 注册 或 登录