These models are mostly suited for linguists and researchers. The output is lowercase and without punctation.