Elk seems to be pushing the RPi as the h/w platform. You will likely have a difficult time finding an RPi with an ethernet chipset capable of correctly doing PTP.

I don't think you need the sub-microsecond resolution PTP compatible PHYs offer. I was hoping to get away with software timestamping which I believe should still give you sub ms synchronization.

My real question, I guess, is how in sync do the speakers need to be? Just on the order of human perception, or do we need them to be in phase?

I have some aging Slim Devices units in my house system that I've been contemplating replacing with a custom system. However, I would just use wifi to an RPi with a suitable DAC installed. There'd be no per-speaker sync, since the DAC->speaker connection is analog, and before the DAC there'd be essentially streaming code that would always be handling both channels.