I can't imagine what my life in the early '70s would have been like without Michael Tearson. As the alpha disc jockey on WMMR, Philadelphia's hippest FM radio station, Tearson spun song sets that seemed to cosmically interact, providing transcendent moments throughout his midnight shift. My pals and I often wondered if he was sending secret messages by his song selection. Only recently, when I finally met Tearson, was the mystery resolved: he was sending us messages--ones of mood, not code--via Beatles, Bowie and Moby Grape.

Personalities like Tearson have all but vanished from the airwaves. But in the era of the iPod and other digital players that store a huge amount of music, there's a rekindling of interest in the art of song sets. Apple's own solution is a double-barreled strategy of handpicked playlists for certain moods and random play for serendipitous discovery. But others are exploring more-sophisticated schemes to algorithmically clone the Michael Tearsons of yesteryear.

This week the Roxio company introduces the Boom Box, a $50 suite of iPod applications that includes a DJ program called MusicMagic Mixer. Originally offered on the Internet, the Mixer is designed specifically to create mood- appropriate yet illuminating combinations of music from your own collection. Its method, according to software architect Wendell Hicken, is to analyze the digital files that store the acoustic information that tells your music player what sound to produce. For instance, by recognizing the bits that encode a drumbeat, the program can divine the volume, tempo and energy of a tune. What's more, by digitally decoding all songs in the same way, it's possible to find hidden affinities between unexpected tunes.

After enduring a long period where MusicMagic painstakingly "fingerprinted" my songs for analysis, I was instantly able to construct some great playlists based on a single "seed" tune that was the keystone of my musical desires that day. Though a seed of alt-country crooner Kathleen Edwards yielded a mix with similar artists like Lucinda Williams and Tift Merritt, it also included an unanticipated but snugly appropriate tune by rocker J Mascis.

A company called MoodLogic takes a different path. Over the past few years it has enlisted thousands of listeners to offer their feelings about specific songs, ultimately collecting millions of comments. Now it uses that "metadata" (MoodLogic VP Christian Pirkner calls it "DJ knowledge") to make playlists that fit a whim. (It plans to license its system to electronics makers.) Like MusicMagic, it will reach into forgotten corners of your collection for the right mix.

Ultimately, we can expect digital DJs to use file analysis combined with metadata, perhaps with more information like song lyrics and personal listening patterns. But even at this primitive stage of robo-DJ'ing, the results can be amazing. Yes, I know that my enjoyment comes largely because the universe of songs I'm working from are all self-selected favorites, and I also know that very special transitions seem more significant because I notice them more than the more common, unremarkable segues. Yet surprisingly often, I get the same satori-esque chills that I did in the days when FM DJs were the oracles of the air.

Back then, of course, the exchange was human to human. "I consciously tried to make each song resonate with the next, which would make each of them more than they would be individually," Tearson told me. "That's really how you bond people with music." How weird it is that now, in our own cones of aural isolation, we may be equally enthralled by an automated selection? No one's sending us messages. But we hear them anyway.