mufin, another online music discovery search engine, went live today. Based out of Germany, mufin takes a unique approach to recommending music.
Unlike its competitors, Pandora, Last.fm, and the new iTunes Genius, to name a few, mufin uses a scientific technology, which analyzes music based on the actual sounds in the song. As described on the Mufin site, “In the essence (also referred to as the fingerprint) all sound
properties of the tracks, such as rhythm, tempo, instrumentation or
sound density as well as additional data, such as the title, are saved… The
advantage is that the origin, genre, popularity or ratings are not used
for the recommendation, only the sound of a music track is decisive
when it is recommended by mufin.”
Its technology is computer automated, whereas other services all involve a human element.
Simply put: Other services recommend music subjectively.
Pandora categorizes music using the Musical Genome Project, which involves human interaction. iTunes categorizes music based on metadata, written into the actual .mp3 file involving a human interaction. Last.fm recommends music based on what other users like and artists they listen to. Once again, human interaction. The list goes on.
I gave mufin a try by searching Miles Davis, “Miles Runs the Voodoo Down”, an experimental Jazz piece. The piece consists heavily of trumpet solos, along with drums, percussion, bass, funky guitar and keyboards.
The results were like none I’ve seen before. One of the top results was Yes, “Heart of the Sunrise,” progressive rock defined and not really in the realm of experimental jazz. With a closer listen, the similar instruments and production style matched Miles Davis. 
The following results made me smile, as I heard all sorts of similar sounds in music styles from Latin Jazz to, believe it or not, Kenny Rogers, “I Will Remember You.”
Although newborn, mufin comes fully equipped with Facebook, Myspace, and iTunes applications.
Unfortunately, for now mufin only allows us to check out short clips of music, but promises to provide full songs in the near future.
Check them out for a different approach to finding ‘similar’ music.
