Musical Motive
Posted on April 22, 2011 by Alison Conard

To begin to explain the many ways that musical understanding has contributed to Clio’s adaptive algorithms, let’s take a look at the smallest identifiable musical unit with its own unique musical identity: the musical motive. Here’s a particularly famous motive from the opening of Beethoven’s 5th Symphony:

Like the Beethoven example here, a motive is a salient, often reoccurring figure that can form the basis of an entire musical work. Motives go a long way toward creating the mood of a particular work. Proof of the power of motives lies in the fact that you only need to hear the first few notes of this example to know where it comes from – it’s a small fragment that identifies the larger work.

Equally powerful is the fact that the motive in a song can change significantly, yet still be recognizable as somehow the same. If we call the opening motive the prototype, we can hear changed versions of that prototype throughout the first movement. Keep listening and count how many times you hear it (hint: it’s a lot):It’s easy for us to hear this motive throughout the movement, but until Clio, computers couldn’t do it.

One of the many smart things Clio does in every single analysis is to recognize the prototypical motive (as well as other motives that occur throughout the piece), catalog its appearances, and calculate each instance’s similarity to the prototype.

Again, the prototypical motive is the bit of music that we remember, the musical idea that identifies the work in our minds. If you know what that motive is and can compare it to motives in other pieces, you can say with confidence, “These two musical works have something really important in common.”

That’s what Clio does. There are, of course, many other things it does well, and when you put all of the pieces together, you have a platform that learns about each piece of music, catalogs the most important bits, and compares them to the important bits in other pieces of music. Sounds a lot like what we humans do, no?

Because its adaptive algorithms are based on human cognition, Clio can perform these analyses on anything we recognize as music. It’s completely style-agnostic – popular music, Balinese gamelan, avant-garde electronic music, and classical music are all equally-delicious fodder for Clio’s pattern analysis.

The “too long; didn’t read” short of it is that the search results, playlists, and recommendations Clio generates are so good because they’re probably the ones that you would generate if you had the prodigious memory of a computer.

Don’t be jealous, though. The computer works for us. At least, for now…