Just a couple of things this week in voice:
> skim The Voice Assistant Benchmark 1.0 results from Cognilytica is out
. The graphics are a bit hard to read because you need to cross-reference them with the definitions of category 0, 1, 2, and 3, but basically these are proxies for how “good” the answer was with 0 being didn’t understand or provided a link to an outside reference, and category 3 meaning it provided a correct answer. I’m most entertained by category 1 where it provides an irrelevant answer. In the graph below, we see that Siri and Cortana have the worst fails, and Google is both the most helpful but also fails about as frequently as it’s helpful.
That’s all for this week, have a great weekend!