Just about all IVR's are coded with a counter for the number of "no inputs" (silences) or "no matches" (whatever you said did not match any of the expected inputs above a configured confidence threshold). So as pointed out above, 6 and 7 are identical, as is plain silence.

Accent detection would actually be pretty hard and beyond the scope of the speech engines used in most of these applications.