The treatment of cogging is either to use a lot of poles, so the impulses are very closely spaced, and there are other tricks of the trade that address the timing, all of which are very effective in rendering cogging inaudible
I have given a plausible (to me at least) explanation of when cogging may become audible even if it is virtually unmeasurable.
Take the Wilson Benesch Table One system as an example. They claim wow and flutter is unmeasurable. The direct drive motor is 15 inches in diameter, weighs 14-kg and has 21 poles, but nevertheless it does exhibit cogging to some degree - I’m not saying it is audible!:
The OMEGA Drive is a slot-less synchronous motor, delivering vanishingly low torque ripple and precise speed stability. The RMS value of the torque ripple is 0.001342 (N.m). The torque ripple includes all the components (like cogging torque and the current commutation and air-gap flux harmonics)

