I just wanted to mention my latest work also in this place. It's called pyfmgraph and it's a visualization on listening customs using the data collected by last.fm.
I need to mention that this is actually not my idea. If seen this before on http://megamu.com/lastfm/ but since this guy didn't reply on my email I decided doing this by myself. While first looking for a convenient program for this task I found none and decided creating my own using Python.
The biggest problem actually was smoothing the curves to have something nice looking. I decided using bezier curves as they are easy to implement and look good.
Now the thickness of each colorstream represents the listening frequency at the time which represents the x axis. Furthermore the colors fade from blue tones to red tones which represents the time the artist has been listened for the first time. Blue means earlier while red references to a later time.
My current work on the pyfmgraph focuses on implementing vector based output since the original application is only able to output rasterized images. If svg output works I maybe could also provide a online version of the program.