Mit model (robinhad#28)

* Use new model

* Black fixes

* Fix examples

* Use GPU when available

* Better handling of downloads

* Bump requirements

* Add link to guide

* Delete LICENSE file

* Add link to datasets

* Add examples

Black fixes

Add version 3.0.0

Release 3.0.0

Add female voice in multispeaker model

Add improved "Mykyta" model with stress support

Add VITS model

Initial model