This work is an application of a synthetic character metaphor to the production of music. Presented here is a small colony of creatures whose virtual bodies produce sound. Played out in a spatial representation, passages of music are exchanged between the creatures and consequently distorted; retrograde variations and repeated subsegments are swapped back and forth in an accidental, recombinant evolution. It draws parallels between musical problems and the problems of motor control and learning in synthetic characters. This work also points towards a system in which characters' movements are shaped (by training) in real time; consequently, by creating music with the resulting system, we are able to perform similar training and shaping within a musical domain.