>"Most models are trained on the original speaker's voice, but maybe only a little bit."
Really cool that you got this to work. I used to work on TTS (a few years ago, now), and we trained on celebrity voices, but used full audiobooks. https://github.com/Kyubyong/tacotron
I think he nixes your case at the start, when he goes through the sanity check thinking about the flat tire. Also, I think your rigid body case use the walls of the tires to apply force to the care, not the force from the pressure to keep the car static, which is why it fails.
Here's an interesting alternative to the current H1B visa. It replaces the lottery with an auction. I imagine this would make companies think twice about hiring "cheap" foreign labor, but then again it might lower the salaries further given the extra burden on the sponsoring company.
From my research, maybe the most interesting characteristic is that all these companies seem dependent on the sucker gripper. I haven't worked in this field, but you'd think it'd be easy getting other gripper to work well, especially since covariant is combining simulated training with non-simulated training.
Really cool that you got this to work. I used to work on TTS (a few years ago, now), and we trained on celebrity voices, but used full audiobooks. https://github.com/Kyubyong/tacotron
Here are some of our Nick Offerman samples: https://soundcloud.com/kyubyong-park/sets/tacotron_nick_215k .