I've created a Processing application called the BPM Explorer that will take a song and display a detailed plot of how the tempo changes over the course of the song.  You can use these 'click plots' to figure out whether a human or a machine is setting the tempo for a song. A machine-generated tempo is, well, machine like.  The BPM for these songs are rock steady  and yield a very flat BPM graph - while songs with human metronomes exhibit much more variation in the tempo over time.  (I've written more about looking at BPM to detect click tracks and drumming machines here: http://musicmachinery.com/2009/03/02/in-search-of-the-click-track/).


The application runs as a Java Webstart app. It lets you analyze any of your own tracks.  Here's a screenshot, that shows the click plot for a Britney Spears track:



while this one shows the click plot for a Nirvana track:

.

The flat line in the Britney plot indicates that a machine was keeping tempo, while the wandering line in the Nirvana plot shows that most likely a human was in charge.

If you are interested in finding out which artists are being ruled by machines you can try out the application: http://static.echonest.com/bpmexp/bpmexp.html

The detailed beat analysis was generated using the Echo Nest API (my employer) using a java client library: http://code.google.com/p/echo-nest-java-api/