Well, with help of Zoltán the original author I've managed to make this work.
He keeps a github repository with a qemu fork and all of his work fully functioning here, although under an older qemu verison.
First fetch a copy of said repository and compile it.
These are my configure options, use "--help" to see what to add or remove:
../configure --python=/usr/bin/python2 --target-list=x86_64-softmmu --audio-drv-list=oss,alsa,sdl,pa
After you've compiled everything you can execute qemu with the following flags:
-audiodev pa,id=asdf,out.mixeng=off -device usb-audio,audiodev=asdf,multi=on
This will expose an usb audio device that supports multichannel audio that gets outputted to pulse audio.
Of note on the previous line, mixeng has to be off for the setup to work, multi "on" to use the multichannel audio device and (the part I had trouble) usb audiodev has to match an id of a declared audiodev ("asdf" in this case).
The author has commented a lot of the options in qemu "--help".
Hopefully someone will pick the work at 2017 GSOC audio work proposal so that it finally gets merged.