It is really amazing!
From the website:
A deep-learning text-to-speech tool for generating voices of various characters. The voices are generated in real time using multiple audio synthesis algorithms and customized neural networks trained on very little available data. This project demonstrates not only a significant reduction in the amount of audio required to realistically clone voices while retaining their affective prosodies, but also the feasibility of an on-demand, stable, and autonomously-improving speech synthesis application that aims to mimic a voice of limited availability. As of January 2021, this current iteration of the algorithm (v11.2.x) is the most cutting-edge in the fields of voice cloning and speech synthesis.