My auditory streaming music sample consists of three big sections split into 10 second increments apiece. The first ten seconds introduces the notes one by one, playing them at an ever increasing speed until three separate voices can be identified. The key components I attribute to the voice split are the differences in loudness and channel direction (both are a large component in all three sections). I also slightly offset the fmIndex, fmRatio, and pitch skew to create a "blow pipe" sound. A descending bass line can be distinctly heard in all sections, and a descending harmony can be heard in the first section. In the second section, two intermingling, warbling voices can be heard distinct from the bass - these are caused by differences in pitch register contour. In the final section, two voices can be heard - a loud, high register one singing the melody accompanied by a two note bass voice.