The use of different state-spaces under different circumstances enables, for example, different state-spaces to be used for different ranges of musical tempo. For example, when no music is detected, a first state-space for a model can be used, when a tempo less than a first threshold value is detected, a second state-space can be used, and when a tempo above the first threshold is detected, a third state-space can be used. This allows for additional synchronization of the pseudorandom animation to different circumstances, and to customize motions (e.g. “dance” moves or combinations of motions from a particular state space) to be customized to different audio data within a single implementation of a computer animation model being executed on a device.