SoLoud supports virtual voices, meaning that you can call play() on many more voices than will be actually audible. SoLoud will then sort the voices by their set volume and play the ones that are potentially most audible.
The default behavior for the inaudible voices is to pause, and resume when audible again. This is fine for ambiance loops, for instance, but is not suitable for some sounds.
If you, for instance, have a gun that shoots every five seconds, having its output end up in the inaudible list will eat up the virtual voices, and when the sounds become audible (by the player moving near the audio source, for example), the sounds will become active at once, which is probably not the desired outcome.
There are also some sounds that need to keep playing even if they should not be audible, because their timing is critical (say, a speech).
You can alter the behavior of a sound source by calling setInaudibleBehavior(), either on an audio source or through the Soloud object for live sounds. Setting both of the flags on is legal, but the sound will just get killed when inaudible.
When gathering the sounds, all voices flagged as "must tick" are selected first. All busses are marked as "must tick" by default. If a lot of voices have the flag on, it may become difficult or even impossible for SoLoud to pick the loudest voices.
You can set the maximum number of active voices via SoLoud's setMaxActiveVoiceCount call:
The higher the number, the more processing power SoLoud will require, but the more concurrent audio streams the user can hear.
To check how many voices are active, you can use the getActiveVoiceCount() function:
unsigned int cacophony = soloud.getActiveVoiceCount();
If the active voice count is constantly at your maximum active voice count, you may have a problem.
Finally, you can check the number of virtual voices in use via getVoiceCount():
unsigned int ghosts = soloud.getVoiceCount();
By default, SoLoud is configured to handle up to 1024 virtual voices. This can be upped to 4095 by editing soloud.h and rebuilding. If even more virtual voices are needed, some refactoring is required.
Copyright©2013-2018 Jari Komppa