This is completely prototype-level at the moment and missing stuff and doesn’t really work, but I think this kind of GUI for the sound sources and the weight/volume snapshots is going to be a lot better than the current solution. (click image for animation)
The rows represent the snapshot states and proportional widths of each sound source block directly represent the probability at which the sound is going to be selected to be played by a fragment. (Which is how it works currently anyway, but is kind of obscured in the old GUI.) This GUI will also intentionally make it more difficult to manage lots of sources, as it emphasizes the sound sources are not meant to be a storage place for candidate sounds. (Which I fear people might have been doing.) If someone represents a convincing sensible case they need to have dozens of sounds loaded etc, I can look into making this work better for such a use case, but initially it might not work ideally for something like that.