I’ve been frustrated by the heavy noise heard in videos captured with the new Sony 24-70 f/2.8 (I know, I should be a better person and use an external microphone and manual focus. But what’s the point of a compact capable DSLR with 4D focus tracking if I need to make it more bulky and give up AF?)

This looks like a great opportunity for a computational audio solution with an extra microphone on the lens to separate the AF noise from the actual audio. One could even imaging doing it solely from the AF control information with a little bit of training and calibration.