Say you see 10 Audubon subs and 4 YRWA species. You also have 6 HOs. As you noted, you can't tell if the HOs are also among the birds you saw. Since you have fewer HOs than sightings, you should assume the ones you heard are among the ones you saw, so report 10 and 4.
What if you'd see 10 subs, 4 species, and had 20 HOs? You heard more birds than you saw, so you should assume the ones you saw also among the ones you heard. You heard 20 total and you saw at least 10 of them were subspecies. You don't know anything about the other 10 you heard, so you can only report them as species and HO. 10 subs, 10 species.
Did that make a darn bit of sense? The basic principle in the eBird guidelines is to err on the conservative side.