That means to technically analyze some audio features like amplitude, power, speed. After the analyzation, all the data were transformed into some basic geometric shapes like linear rectangles, ellipses or triangles. Those shapes keep rotating, iterating, spinning up into a circular spectrum while the speech is playing and this process is also manipulated by the sound parameters. So I generated a kind of rolling vortex to reflect the emotions of this speech, which consists of complex graphical geometries.