The question of effectively communicating an artwork in a cultural context relies on joint understanding of certain creative conventions that are shared between an artist and his audience. Modeling of such cultural entrainment requires representation of a style that is specific to each genre, a task that depends in turn on particular compositional rules and aesthetic sensibilities of each culture. In this chapter we extend our previous research on machine learning of musical style into a broader approach of modeling aesthetic communication. The underlying cognitive assumption of our model is that listener’s experience of music is a process of actively seeking explanation by reducing the complexity of an incoming stream of sound through a process of approximation and prediction. Musical Information Dynamic is an analysis method that measures changes in the amount of information contents of musical signal over time. Motivated by semiotic analysis, we apply information dynamics analysis in order to measure the tradeoff between accuracy or level of approximation of a signal as captured by its basic units, and its overall information contents derived from its repetition structure. This approach allows us to formally analyze cultural communication in terms of aesthetic and poietic levels in paradigmatic analysis. Comparisons of flute recordings from Western and Far Eastern cultures show that optimal sensibilities to acoustic nuances that maximize the amount of information carried through larger structural elements in music are culture dependent.