How UnityEngine Speech works (understaing it's capabilities)

I'm experimenting with voice control. I used this tutorial and it works fine. English words work. But not some other language words (Slovak) or made-up words. So I ask how does it work, does it have its own library of words? I've heard it uses phenomes so basically you just need to say the thing and it should work no? IDK. If it works with a finite amount of words - is there a way to teach him new words? Or is there any other API (or asset) easy to use? (with custom words, or learning possibility)