Exploring the hidden potential of sound data