I wonder what the legal implications of this alongside similar developments like deepfakes are going to be in the next couple years. We're already having fraudsters impersonate CEOs using Deep learning-aided Voice generation[1] due to just how low the barrier of entry is now. There's already a public implementation of the paper out [2]!
CorentinJ's implementation isn't quite as good as Google's - I think with some of Google's samples I couldn't tell that they weren't real, especially over the phone. But I could easily tell with CorentinJ's.
That seems to be common with open implementations of Google's voice synthesis and speech recognition work. I guess they hold back some of the secret sauce, or can afford to train it more.
[1]: https://www.wsj.com/articles/fraudsters-use-ai-to-mimic-ceos... [2]: https://github.com/CorentinJ/Real-Time-Voice-Cloning