A Radio Station Just for You
FOR all the talk about satellite radio, the most vibrant frontier in radio may be the Web. Many traditional AM and FM stations have begun streaming on the Internet, along with hundreds of smaller online-only operators. Even subscription download services like Napster, Rhapsody and Urge from MTV have preprogrammed radio for users who are not in the mood to hunt for tracks.
Currently, the most compelling online radio is interactive. Services like Pandora, Last.fm and Slacker evaluate your musical tastes, then serve up a continuous stream of programming to match. They mix familiar songs with new material you might like. They all do it by harnessing the technological forces of social networking, data mining and music analysis, though each uses a slightly different technique.
With so much momentum, there are still plenty of bumps. The Copyright Royalty Board of the Library of Congress recently announced a Web-radio royalty payment plan that has caused many free Internet broadcasters to fear for their fragile business models.
Some new interactive music services choose not to stream anything. Instead, they rely strictly on music the listener already owns or new tracks donated by publicity-hungry independent artists and labels. Others are becoming as creative with the way they license content as they are with the way they personalize it for you.
On its surface, Pandora is the simplest option. When you visit www.pandora.com, enter the name of a song or artist you enjoy. Immediately you will hear music from a “station" based on that initial choice. You can refine your station by naming other artists and songs, and Pandora picks music from those artists but more important, it chooses other songs you might like based on your suggestions.
Pandora makes recommendations based on analysis of songs by musicologists in Pandora's Music Genome Project. The experts listen for up to 400 different characteristics in every song, from musical genre to the presence of a particular instrument. Songs with the most similarities naturally make their way to the same radio stream.
If you do not like one of Pandora's suggestions, you can click on the “thumbs down" sign and it is never heard from again. If, on the other hand, you do like a song, you can give it the “thumbs up," and that particular preference will be used in later suggestions.
Now that the free ad-supported service has been operational for 15 months, it can use the behavioral data of its six million listeners to add a new layer of suggestion. For instance, even if, on paper, the musicologists think it logical to pair a song by the “American Idol" superstar Clay Aiken with one by the Canadian folk balladeer Ron Sexsmith, several hundred listeners may give the juxtaposition a vote of no confidence. Tim Westergren, a Pandora co-founder, says the database now contains half a billion useful points of “contextual feedback."
Continue...
Currently, the most compelling online radio is interactive. Services like Pandora, Last.fm and Slacker evaluate your musical tastes, then serve up a continuous stream of programming to match. They mix familiar songs with new material you might like. They all do it by harnessing the technological forces of social networking, data mining and music analysis, though each uses a slightly different technique.
With so much momentum, there are still plenty of bumps. The Copyright Royalty Board of the Library of Congress recently announced a Web-radio royalty payment plan that has caused many free Internet broadcasters to fear for their fragile business models.
Some new interactive music services choose not to stream anything. Instead, they rely strictly on music the listener already owns or new tracks donated by publicity-hungry independent artists and labels. Others are becoming as creative with the way they license content as they are with the way they personalize it for you.
On its surface, Pandora is the simplest option. When you visit www.pandora.com, enter the name of a song or artist you enjoy. Immediately you will hear music from a “station" based on that initial choice. You can refine your station by naming other artists and songs, and Pandora picks music from those artists but more important, it chooses other songs you might like based on your suggestions.
Pandora makes recommendations based on analysis of songs by musicologists in Pandora's Music Genome Project. The experts listen for up to 400 different characteristics in every song, from musical genre to the presence of a particular instrument. Songs with the most similarities naturally make their way to the same radio stream.
If you do not like one of Pandora's suggestions, you can click on the “thumbs down" sign and it is never heard from again. If, on the other hand, you do like a song, you can give it the “thumbs up," and that particular preference will be used in later suggestions.
Now that the free ad-supported service has been operational for 15 months, it can use the behavioral data of its six million listeners to add a new layer of suggestion. For instance, even if, on paper, the musicologists think it logical to pair a song by the “American Idol" superstar Clay Aiken with one by the Canadian folk balladeer Ron Sexsmith, several hundred listeners may give the juxtaposition a vote of no confidence. Tim Westergren, a Pandora co-founder, says the database now contains half a billion useful points of “contextual feedback."
Continue...
For more information contact All About Jazz.
