Leonardo Music Journal, Vol. 3, 1993, pp. 45-52.
Balance can be a primary concern in computer-generated art and music. Transforms used to obtain statistical balances are typically built on the assumption that the sequences driving the transform range uniformly from zero to unity. However, few sequence generators are uniform over the short term. The author describes a method, called leveling, that transforms an arbitrary, nonuniform sequence so that the sequence retains its essential shape while its distribution becomes uniform from zero to unity. Applications of the author's leveling algorithm to Brownian, chaotic, and fractured sequences are explored. The conclusion compares leveling with other rigorous balancing methods including statistical feedback.