I'm not actually sure how I feel about this. Australian English is distinct, and one of its key features is heavy use of shortened words and dropped letters. For example, the word "going" becomes "goin".
My intuition (and complete guess) is that ML is pretty far away from distinguishing slang words. Maybe it can understand it, at least to some extent, but categorizing it by country (well you really even need region) is harder.
I'm Australian, and I would never recommend that someone learn English by listening to one of my countrymen. We are lazy, nasal, slurred speakers. Standard English or American are much more preferable.