Software Picks Music That Suits Users' Tastes
February 25, 2008 at 5:29 AM (PT)
Software that predicts the music you will like is behind the dreams of a CHAMPAIGN-based start-up company that hopes to boost tunes from independent bands, reports THE CHICAGO TRIBUNE. The company, ONE LLAMA, has launched a website that strives to make playlists of songs based on a user's tastes. Once a user enters tunes he finds pleasing, the computer searches for others that it expects will also find favor.
It's a simple concept backed by complex software that grew out of two decades of acoustical research by UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN computer geeks who are also musicians.
"We use two systems," said AMIT SUDHARSHAN, who plays the piano and violin and who recently received a computer engineering degree. "The first system is cultural analysis: How often are songs played on the radio with each other and placed on playlists with each other? We look at pop culture as reflected on the Web. The second part is based on the acoustical quality of the songs. The computer hears two songs and judges their acoustical similarity."
Software quality is the key to success, and the fledgling firm has sold a commercial version of its software to EMI MUSIC, enabling people to search the EMI catalog.