The configuration request now lists OBS's mixer tracks and their names (Settings → Output → Audio), so a server can ask for named tracks as stems. Every audio track the plugin encodes from the server's configuration is written with tkhd.alternate_group = mixer track + 1, so the AAC and Opus encodes of one mix share a group, and with the mixer track's name as its title in trak/udta/name, where ffmpeg writes one. Reading the names needs obs_frontend_get_profile_config, so the build pulls in the frontend API header and, on Windows, an import library for obs-frontend-api.dll.