Scientists increasingly rely on computational models of atoms and molecules to observe, understand and make predictions about the microscopic world. Atoms and molecules are in constant motion, with vibrations and structural fluctuations occurring at very short time-scales and corresponding length-scales. But can these microscopic oscillations be converted into sound? And, what would they sound like? In this paper we present our initial steps towards a generalised approach for sonifying data produced by a real-time molecular dynamics simulation. The approach uses scanned synthesis to translate real-time geometric simulation data into audio. The process is embedded within a stand alone application as well as a variety of audio plugin formats to enable the process to be used as an audio synthesis method for music making. We review the relevant background literature before providing an overview of our system. Simulations of three molecules are then considered: 17-alanine, graphene and a carbon nanotube. Four examples are then provided demonstrating how the technique maps molecular features and parameters onto the auditory character of the resulting sound. A case study is then provided in which the sonification/synthesis method is used within a musical composition.