That could be -- there are more people whose genders are not as they seem than most realize. But they're surely not going on voice alone but also name, in-pers

That could be -- there are more people whose genders are not as they seem than most realize. But they're surely not going on voice alone but also name, in-person impression based on appearance and so on. I once made a gender survey of several hundred people going by names, falling back to pictures in LinkedIn profiles, falling back to pronouns in forum posts referring to the person in question by people who ought to know. I assume the results were not perfect -- but they were probably good enough to be useful. (The result was 95+% male, which was enough to prove a point.)