Hey Siri: “the computation in the “Hey, Siri” detector is in the acoustic model.” Posted on October 18, 2017 by jgordon Link. On second attempt more sensitive
