Apple didn't really talk about it directly today, but I'm wondering what level of back-end improvements Siri will have to improve recognition while issuing commands or dictating. It does sound from today like a lot more will be happening on-device. Anyone at WWDC getting insight on this?