Most of the information in our brains is conveyed by discrete events (action potentials) that can be interpreted as a digital code. However, our senses receive analog inputs fromcells generate graded potentials in response to stimuli. But their primary nerve afferents re-encode this information as a temporal pattern of action potentials. In auditory, visual and vestibular systems a highly specialized synapse (ribbon synapse) plays a chief role in such a conversion. The ribbon synapse has unique features: it can release neurotransmitter at high rates for sustained periods, it is extremely fast and reliable and it has little or no plastic behavior. It is thought to achieve this by means of coordinated multivesicular fusion; but the way it does remain unknown. In this work we propose a dynamical model of the ribbon synapse that is able to reproduce the main features of these units. Un- like previously reported models, the dynamics of the vesicles of neurotransmitter is strongly coupled. We use our model to compare two recently proposed underlying mechanisms of multi-vesicular release: compound fusion and coordinated re- lease. For the case of the auditory system, we also show that these synapses are crucial to encode temporal information in the sub-millisecond range, a requirement for the localization of sound sources and the perception of pitch.