Why do localised voices in events using auto-defined soundbanks stop playing when I call SetCurrentAudioCultureAsync?
I'm writing a locale test using Unreal 5.4 and Wwise 2022.1.18. We're using user-defined soundbanks for localised voices. The reference language is English(US) (en-us).
The Unreal -> Wwise language mappings have been set up as described here:
https://www.audiokinetic.com/en/public-library/2024.1.4_8780/?source=UE4&id=using_features_localization.html
The soundbanks are generate correctly and a .wem exists in each localised folder. The uasset representing the event exists in the editor and can be heard in English when right clicking and selecting "Play Event".
A blueprint cycles to the next supported Unreal culture, calls SetCurrentAudioCultureAsync() and then posts the event that plays the localised voice.
If we never make the call to SetCurrentAudioCultureAsync() the event plays audibly in English.
When SetCurrentAudioCultureAsync() is called, the event fails to play. The success code returned from UAkAudioEvent::PostEvent() is AK_INVALID_PLAYING_ID (0). It fails on this check:
if (!IsDataFullyLoaded())
{
UE_LOG(LogAkAudio, Warning, TEXT("Failed to post AkAudioEvent: Not all localization data for '%s' are loaded. Consider using PostEventAsync()."), *GetName());
return AK_INVALID_PLAYING_ID;
}
My understanding is that auto-defined events are loaded as needed. Since we stop all events prior to changing the culture all referenced banks should be unloaded. Afterwards the localised versions of these events should be available to play. I've read all the documentation stating that soundbanks need to be unloaded before performing a locale switch, but there was nothing that specifically referred to auto-defined soundbanks. AFAIK there is no manual load/unload for them. Have I missed a step somewhere? And why would this work prior to calling SetCurrentAudioCultureAsync()?