TL;DR

I woke up today and thought, hmmm, my AI needs a voice. He can’t just be a moron generating silly text like a disabled ape. Voice gives more character to an otherwise “computer-like” entity. Below are voice samples for AI Youcef which I generated using various voice cloning Machine Learning methods. AI Youcef can now speak to humans. and speak he does, more than 10 different Languages fluently. This is one step further to reach digital immortality I guess (this was a joke, no need to take it seriously).
Completing this with voice recognition is going to be demanding though, I’m gonna need resources, but it can be done easily.

Note: All audio tracks were generated only from TEXT. I just made a TTS model learn my voice from a simple original recording and that was all it needed to extrapolate how my voice would sound in other languages.

Lastly, this beauty

TECHNICAL BS and Credits

I used tools and resources from the coqui-TTS Github repo.
The Technology stack for this rather small endeavour is amazingly huge, PyTorch, Cuda, Numpy, scipy, scikit-learn, and other +55 dependencies in total. So most credit goes to the devs who open sourced the tools for us mortals to implement.