That's the question. The article I've found is from 2016. Maybe (all) modern phones handle that properly.
Captured a few seconds while a vowifi call was active. At least on my phone, Google Pixel 9pro running Android 17, the DSCP EF is applied on the egress ESP frames.
FWIW I played with this in context of WMM and VLAN and discovered that manual mangling was needed via bridge's IP firewall. At least previously and perhaps was only necessary for my network layout. Mikrotik has a relevant article: WMM and VLAN priority - RouterOS - MikroTik Documentation
But if you don't use VLANs in that segment you can disregard this.