Sun Microsystems has unveiled a technology that will recommend songs to users based on an analysis of the music that they already enjoy.

The technology analyses features such as rhythm and beat strength to categorise the music. It then searches for files with similar attributes.

Sun showed off the 'Search Inside The Music' project at Sun Labs Day, an open house for its research arm in Mountain View, California.

The system takes about three seconds to analyse one minute of music, explained Paul Lamere, a principal investigator with Sun Labs.

Because the analysis is such a compute-intensive task, he expects that users will acquire the data from a server or have it bundled with the song at the time of purchase. The analysis is further augmented by metadata concerning the tracks' genre.

The tool will help users find music they like on the next-generation digital music players, which Sun expects will offer storage capacity for one million songs.

Lamere predicted that future devices will hold music files not just for the owner, but for other people such as partners and children. This requires a new technology to allow the user to listen only to the music he or she likes.

The product also lets consumers discover new music. "There is currently no avenue for new music to get into the systems. This gives a way for garage bands to come in. It's putting all the songs on the same footing," Lamere told vnunet.com.

'Search Inside The Music' is currently a research project, and Sun does not yet have any plans to turn it into a commercial product.