I scored very well (82% for faces, 92% for voices), which is surprising since I don't think I read people terribly well. It does make me think, though; I think my difficulty with socializing isn't necessarily reading people's faces or voices, but rather intuiting ToM--what is a person likely thinking? How do they feel about a certain situation? How will they perceive what I say? I might be able to read their face for "big" emotions, but I don't reflexively jump into people's heads and see 'oh, they're going to feel XYZ because of ABC,' so I'm not on the look-out for subtle things.
In any case, I don't think this test was very well-designed.
The voice one would have been better had they made the actors say phrases that didn't necessarily match up to the emotion they were trying to say. If someone says something like, "Oh, god, this is terrible!", you can rule out that they're not "vibrant" or "reassured." I wasn't always sure of the answer based on the tone itself, but I was able to successfully intuit it from the phrase's content alone.
I also don't think people's facial expressions last for that long or are that exaggerated. If you say something that annoys other people, chances are, their mouth isn't agape and their eyebrows furrowed for two full seconds. Rather, they're more often microexpressions that last for just a fleeting second.