2 ms·

I tried that. But their recognition model seems to be quite picky. And you need the voice change real time. No time to edit anything.