[modeless]

I expect that the training data for the voice AI is overwhelmingly English. Native speakers of other languages will be a small minority. I'm sure it will improve over time.