I'm building various prototypes for VR training simulations using Inworld. But they also use the cascaded approach. Also, I am building customer service agent product which we would love to add voice to but whisper and eleven labs (and others) are just too slow. Is tincan available via API?