"How", is very dependent on your equipment. From the tags you mention, I would guess they are DSCP markings, with the VoIP ones being set by the VoIP devices. Often L3 ToS is passed end-to-end unless a L3 device, or devices, modify it. Also often, L3 devices might not, by default, provide any different handling of such tagged traffic.
At the L2 level, if the device supports it, you often need to work with L2 CoS, often mapping up/down from L3 ToS. L2 devices often have to be explicitly configured to support CoS and also configured what to do with the markings. (L2 features are often much less capable than L3 features.)
In general you should consult the documentation for your devices to see what they support with regard to QoS with the goal to send queued outbound voice bearer packets/frames first, insure voice signally packets/frames are neither dropped nor delayed too much, and treat all other traffic "best effort".