In the system we deliver, there is a long-lived communication application and we hope it won't affect the other audio playout in system (no need for ducking experience at all), so we change the stream attenuation setting from "80% lower" to "disable" and it takes effect immediatly.
Then we found that the setting won't keep working when system reboots, event the GUI setting is still on the disabled state -- it will work again if we mannually change it to other options and then change it back to disabled.
Currently we uses WIN10 Ent 2019 LTSC (1809), don't know if it's been fixed in the later non-LTSC release.
Similar problem has been reported in this avsim thread, looks like manual setting change will trigger some background operations other than the pure registry value (UserDuckingPreference) update.
Also, we have tried to use the method introduced in this win32 api doc to avoid the stream attenuation for a specified application developed by ourself, looks like the attenuation still happens to it (not as much as 80%, maybe around 50%).
Any advice/suggestion would be greatly appreciated.