StatBalProc:
Filter:
BBCut/Classes (extension) | Libraries > BBCut > Cut procedures

StatBalProc
ExtensionExtension

Automatic breakbeat cutting algorithm using statistical feedback

Description

Extension of ChooseCutProc using Charles Ames' method of statistical feedback to control selection of cut size and repeat states. The essential auxilliary class is StatBalStream. The theoretical basis of the use of statistical feedback is control over a continuum from Lehmer style random number generation to deterministic behaviour via the heterogeneity parameter. The application to automatic cutting is to promote different types of short term diversity with long term proportions preserved for parameter choices. Note that statistics are slightly perverted by phrases, which may require that only a smaller cutsize or number of repeats fits in the available space. Set phraselength very large to avoid that.

It is fair to say that the effect can remain unnoticeable, depending on choices of heterogeneity. Low heterogeneity statistical feedback restricts the choice of potential cut sequences, avoiding strange dispersions of states (that is, rare states possibly cropping up many times in a row).

For additional data on the algorithm see Nick Collins, "Further Automatic BreakBeat Cutting Methods", Proceedings of Generative Art 2001, reproduced at http://www.cus.cam.ac.uk/~nc272/.

Class Methods

.new

Arguments:

cutsbs

A StatBalStream object that returns a cut size from the next method.

repeatsbs

A StatBalStream object that returns the number of repeats from the next method.

phraselength

Next length of phrase in beats.

rollchance

A probability of a roll near the end of a phrase.

rollallowed

number of beats within sight of the end of the current phrase within which a roll is permissible.

bpsd

beats per sub division. Sets a primitive cut size resolution for choose offset messages.

Examples

You should substitute your own break sample to hear the effects of this cut procedure more clearly. Either way, the effect is fairly subtle.