Anyone who loves music knows that the Internet was probably the greatest thing since the invention of the Sony Walkman (or the iPod, for a younger demographic). And not just because you can find mp3 files of your favourite songs -- either legally or illegally -- but because you can also usually find the lyrics, the musical notation, CD cover art, band history, tour schedule and dozens of websites put together by obsessed fans, all with the click of a mouse.
As the Web has evolved over the past year or so, with more sites and services incorporating the interactivity and community design some call "Web 2.0," the kinds of things available to music fans have also evolved. Now, instead of trying to find new music by trolling through comments on a bulletin board or relying on tips from friends, you can try out a "recommendation engine" that take your musical tastes as a starting point and suggests new artists or songs you might like. There's a wiki-powered "Music 2.0" directory (which anyone can contribute to) in progress here.
In a sense, these services are like your own private radio station -- one which knows enough to play what you like, but also throws in the occasional new song, and lets you rate it. The idea is that such software "learns" about your preferences over time. You can also share your preferences with others, who can incorporate your station into theirs. Companies have been working on this kind of "artificial intelligence" for years, but early versions were unimpressive. Some of the more recent attempts, however, are starting to gain fans.
Two of the most popular sites or services in this category are Last.fm and Pandora. Last.fm was started in London in 2003 by former music journalist Martin Stiksel and a couple of fellow music fans, and now gets about a million unique visitors a month. The site has a library of more than one million songs that it is licensed to stream to users. While that is smaller than the 3.5 million songs that Apple has available through its popular iTunes music store, Last.fm is designed primarily for listening as opposed to buying.
When you first get to Last.fm, it asks you to name an artist you like or choose a genre, then it starts streaming some songs through a pop-up flash player and asks you to rate them by clicking on various buttons (a heart icon means you like it, a circle with a slash through it means you hate it). You can tag songs with key words, or you can recommend them to a friend. You can also download software that takes what you listen to in your iTunes or Windows Media Player and uses that to help determine your preferences.
Whereas Last.fm sorts and recommends music based on community voting -- suggesting new songs for you based on who else liked the same songs you liked -- Pandora uses software to accomplish the same thing, by using a song's musical "fingerprint" to compare it with others and generate suggestions for you (it's based on something called the Music Genome Project). The company employs dozens of sound editors and musicians who categorize music into different genres and sub-genres, based on the type of instrument being played, the melody, the rhythm, the artist and so on. When you select a song, you can see all of that information in the player.
While listening to a song in the embedded flash player, you can click a menu and say whether you like a song or not, and you can also choose buttons that say "I'm tired of this song, don't play it for a month" and "Why is this song playing?" The latter brings up a a panel that tells you the attributes the software chose the song based on, such as "acoustic rock instrumentation, vocal-centric aesthetic and major-key tonality." From a menu in the player, you can also bookmark a song, tag it with key words or buy it through iTunes.
If you look around the Web, you will find ardent supporters and champions of both Last.fm and Pandora, with both sides arguing that their method -- community vs. smart algorithms -- does a better job of finding music you might like. The best advice is probably to try both services for awhile and then see which one is better at giving you what you want.
Other sites include Yahoo's LaunchCast (which also does music recommendation) as well as Live365.com, which has about 7,500 user-generated music stations you can choose from, and MusicStrands.com, which is another music recommendation engine and community-oriented site. Goombah, music software that analyzes you iTunes or Windows Media Player lists and then suggests other artists based on community ratings, just launched recently, and features an "adventurousness" slider that lets you increase or decrease the amount of difference between what you like and music you've never heard before.
If you feel like just wandering around the Internet listening to what some other people are listening to -- without any real organization or categorization -- there's an interesting site called The Hype Machine that streams random mp3 files from various music-related blogs. And for something completely different, there's LivePlasma, which also groups artists based on their musical similarities, but does so in a cool, graphical format with coloured spheres representing the artists, which reorganize themselves when you click on a particular sphere. Maybe not that useful, but lots of fun to play with :-)
