Extend this for other languages

Hey, thanks for this repo. I am currently working on Cantonese GOP scoring proj. And since there are no open scoring datasets available like Speechocean , what would you suggest ? How the scoring model be trained and later evaluated for GOP ?