OK, I recommend temporarily removing Rejecto, turning the vadThreshold up to 4.5 and reducing it by increments of .1 until you find the highest value which perceives your word and doesn’t react to the music. Once this is established, re-add Rejecto to reject OOV human speech.