[darkvertex]

What do you use for validating pronunciation?

Or do you just use a Speech To Text model and hope the text that comes of it is the expected phrase?