Voices may become virtual when their volume goes under threshold. For any given sound, the volume that is compared against the threshold is the real effective volume of all its audio channels, resulting from the contribution of all volumes of the actor-mixer hierarchy, fade transitions, interactive-music transitions, RTPC, States, positioning and attenuations.
The contribution of fade transitions also counts when computing the effective volume of a sound! Consequently, during a crossfade transition inside a random-sequence container or a blend container, fading sounds will likely go below volume threshold for some time. If their behavior under threshold is "From Beginning" or "Resume", their real duration will be longer than expected by the container's logic. This will result in unpredictable behavior. Worse, when a sound fades out under threshold, it stops "virtually" playing, and therefore never ends, so the container might stop sequencing more sounds.
To avoid these issues, you should use "From Elapsed Time" virtual voices with containers with crossfade transitions.
