A recently-granted patent application could point the way towards the next-gen HomePod having facial recognition and control via hand gestures.
As with all such patents that appear online after they lose confidential status, it was originally submitted a while ago - back in 2017 in this case. And while the HomePod isn't explicitly mentioned, it's pretty clear that that's what the patent is referring to.
The lengthy document talks about "a voice-controlled assistant device (e.g., a countertop speaker)" which it refers to as device 10.
It mentions a huge amount of potential sensors and scenarios even mentioning that the Siri icon (the patent refers to a generic digital assistant) could be a screen that changes to show an emoji reflecting your mood: "If, for example, sadness is detected in the user's voice, the avatar may change accordingly (e.g., to a sad emoji to represent user sadness or to a happy emoji to counteract the detected sadness)".
There's a huge hint at Face ID support in the device: "User proximity to device 10 may be measured using proximity sensors or other equipment. If desired, the identities of individuals in the vicinity of [the device] can be determined using facial recognition (e.g., by performing image processing operations on images captured with cameras in device 10)".
So there's a clear reference to proximity as with recognition - could HomePod 2 therefore be able to recognise different family members?
Most notably the patent talks about "sensors that gather hand gestures and other three-dimensional gesture input (e.g., using optical sensing, capacitive sensing, etc.)".
As MacRumors points out, this could utilise tech Apple bought in 2013 as part of 3D sensor startup PrimeSense which was also involved with early versions of Microsoft's Kinect sensor. It obviously thought it was worth buying; reports at the time suggested the price was set at around $360 million.