People often ask me what music I listen to, but I find it difficult to describe my eclectic taste in words. 

I can say that I like Thelonius Monk and Ella Fitzgerald and Béla Fleck and Beethoven. I do like music by these musicians, but they present a distorted view of my taste. They are simply the names most likely to be recognized. 

I can also say that I like Eleanor Hovda and Faradj Karajev and Michael Gordon and Josquin. I do like music by these musicians, but most people have never heard of them. They make me seem like an aloof, academic snob. 

I can also say that I like quiet, sparse music, that I like conceptually-driven work, that I appreciate clever allusions and thoughtful references. I do like all of these things, but it is difficult to extrapolate my musical tastes from such vague descriptives. 

So instead of answering this question with words, I now answer it with sound. In 2005, I developed a web-based software program titled iTunes Signature Maker (iTSM). It analyzes the music in a user’s iTunes music library, both in terms of her listening habits (play counts, star ratings, play dates) and the audio content of the files themselves (spectra). It uses this information to quickly generate a short audio file that mixes together segments of the user’s favorite songs to create a concise portrait of her musical taste in sound. 

The software is not perfect by any means. Its algorithms are crude and simplistic, because I wanted them to execute in a few short minutes. (Who would be willing to let their computer churn away for days on end just to generate an iTunes signature?) It relies heavily on personal listening data tracked by iTunes, but this data can miss important details of listening habits. (If someone has played a song a thousand times, iTunes does not know whether it was played mostly last week or mostly many years ago.) And it cannot include songs purchased through the iTunes Music Store that are protected by digital rights management (DRM). 

iTSM’s worst signatures resemble advertisements for a radio station or a Time-Life music compilation: short snippets of songs quickly cross-faded together. iTSM’s best signatures are themselves interesting sonic objects, with individual fragments linking together to form larger musical phrases. 

Even though iTSM’s signatures are not perfect, they are still much better than words at describing my musical tastes. In fact, sometimes I wish they were less accurate: the inclusion of some musical works in my signature embarrasses me. I am proud that my signatures often include music by Charles Ives. I am not as proud that they often include covers of Leroy Anderson’s holiday classic, “Sleigh Ride.” But in the end, my signatures instigate self-acceptance: all of this music is an important part of who I am, and all of it is beautiful to me. 

I intended iTunes signatures, of course, to be mechanisms for both self-reflection and for self-expression. In the spirit of the latter, I encourage users to share their signatures in an online gallery, to post them on their weblogs and web sites, and to discuss them with each other. One of my own signatures serves as my mobile phone ringtone. These signatures, ultimately, are part of larger cultural trends toward building online identities around listening preferences. 

The five tracks presented here, collectively titled “What I Listen To,” express my eclectic musical tastes. It is a series of short experimental sound works that algorithmically stitch together bits and pieces of the music I've listened to. I created them during the development of the iTSM software; each provides a different snapshot of who I am and what I listen to. The algorithm considered both my listening habits and the spectral content of the sound files themselves when creating these pieces. The results are a mix of smooth textures, chaotic collages, and embarrassing revelations about my taste in music.