Implementation used an unofficial (but popular) Implementation of the famous paper "Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis "

This demo is used only for educational perposes.

Results:

Reference voice 1:

https://user-images.githubusercontent.com/57376705/137780166-860e95f1-8b69-44c5-8728-24658a380752.mp4

-Hi people check this out this man Rida is doing something fun

https://user-images.githubusercontent.com/57376705/137780890-b5533efb-cbd1-4348-a49c-ee22bb7e701c.mp4

-Morocco is a beautiful country that you should visit at least once in your life

https://user-images.githubusercontent.com/57376705/137781954-2f4695e1-fb24-49dc-b52c-4bf35a31d608.mp4

Reference voice 2 (Barack Obama):

https://user-images.githubusercontent.com/57376705/137780154-dd4e6fe2-2251-4b52-a934-77deb14395d9.mp4

-Hi people check this out this man Rida is doing something fun

https://user-images.githubusercontent.com/57376705/137780149-76a25897-7a32-430c-bc72-7118addaaf61.mp4

-Morocco is a beautiful country that you should visit at least once in your life

https://user-images.githubusercontent.com/57376705/137781957-7c8093ba-4db8-4763-9cea-0b9823683b7c.mp4

References:

Implementation using the model from: https://github.com/CorentinJ/Real-Time-Voice-Cloning

Based on the paper: https://arxiv.org/abs/1806.04558