It allows for the separation of transient and stable components of a sound by doing an analysis in search of wave forms that change phase at faster (transient) or slower (stable) rates. 

The extractor actually divides the sound up into two sound files, one containing the transient elements of the original and the second containing the stable elements.