Start-up composes a 'music genome'

OAKLAND -- In a downtown office building an hour's drive north of Silicon Valley, the technology revolution beats on.

Young musicians, their headphones plugged into desktop computers, are analyzing thousands of songs -- from popular artists to garage bands -- by more than 400 musical measures. In an adjoining room lined with rock posters and shelves crammed with compact discs, T-shirted engineers are shooting pool while employees on their lunch break jam in an acoustic rock session.

Welcome to Savage Beast Technologies Inc., a five-year-old software start-up that is busy building a ''music genome" to identify the world's recorded music according to vocal, lyrical, melodic, harmonic, and instrumental attributes. The results are fed into a massive database to spit out recommendations for music-loving shoppers at Best Buy, Borders, AOL, and other retailers that license Savage Beast technology.

''I'd like our genome to become the standard, to become part of the lexicon of musical discovery," said Stanford-trained pianist Tim Westergren, 39, founder and chief strategy officer of Savage Beast, who formerly played in rock bands. ''I'd like people to think of it as a trusted source to give them the best enjoyment of music."

Savage Beast is emblematic of a new generation of Bay Area start-ups, fueled by venture capital, that incorporate technology into more traditional fields, from music, media, and entertainment, to retail, financial services, and consumer products. At a time when businesses are spending more cautiously on high-tech equipment and services -- the sweet spot of the 1990s technology boom -- venture capitalists are more willing to place their bets on companies like Savage Beast that seek to innovate on the intersection of technology and entertainment.

The company is part of a small cluster pioneering so-called ''recommendation technologies" that seek to connect music buyers of varying tastes to songs and artists with which they might be unfamiliar -- just as online bookseller Amazon.com does with book titles. Other companies in the space include ChoiceStream Inc. of Cambridge (which also recommends videos and television shows), Yahoo's Launch service (formerly Musicmatch), and the Spanish-based Polyphonic HMI.

''Most of these recommendation technologies are hit-and-miss right now," said Josh Bernoff, principal analyst at Forrester Research in Cambridge. ''If a service can get it right nine out of 10 times, that's useful. If they can do it five times out of 10, that's just annoying."

Unlike competitors that use ''collaborative filters," linking users to favorites from others with similar tastes, Savage Beast seeks to base its recommendations on a more scientific method -- breaking down songs by the depth of their bass, the breathiness of their vocals, and myriad other features. If you like a song by the Rolling Stones, another by Dire Straits, and a third by Van Morrison, plugging their names into Savage Beast's database will yield a list of songs -- many from lesser-known or independent bands -- with comparable musical attributes.

Consumers can access the database today only through retail channels. Savage Beast technology currently powers kiosks at some Best Buy and Borders retail stores, mostly on the West Coast.

One of the main tasks of Savage Beast's new president and chief executive, Joe Kennedy, a 45-year-old auto industry veteran who headed marketing for Saturn, is to figure out how to exploit the exploding digital media market in the era of the iPod.

''To me, music is an industry that's in complete transition," Kennedy said. ''The key problem is to connect people with just the music they like."

Founder Westergren, who spent years on the road playing keyboard for YellowWood Junction and other bands, began thinking about that problem in the late 1990s when he was writing musical scores for films. Directors would give him the name of songs they liked and ask him to compose something similar. Westergren became convinced that, as music became global and digital, there would be far too many choices to approach that way. He was also miffed that music labels weren't taking chances on independent artists who hadn't attracted large audiences, and he hoped to find a new platform to get their music to mar-ket.

Savage Beast, launched in 2000, raised $1.5 million in seed funding from Garage Technology Ventures, a Palo Alto, Calif., firm led by former Apple evangelist Guy Kawasaki. But then the technology bubble burst, and the door slammed on venture funding.

''We barely survived," Westergren said. ''We ran out of money 50 times in the next three years."

The start-up moved from San Francisco to cheaper quarters in Oakland, and staffers took voluntary pay cuts.

But many stuck it out, working by day as musician-analysts and playing in bands at night. Savage Beast analysts get 40 hours of training before they begin classifying music, with trainers that include musicologist Nolan Gasser, a keyboardist for the Steve Miller Band.

Last year, the company received a second round of funding totaling $7.8 million. The round was led by WaldenVC of San Francisco and included Labrador Ventures of Palo Alto.

''At the core is the notion that people love music and that discovering music is a big problem," said Larry Marcus, a WaldenVC general partner who sits on Savage Beast's board. ''They've come up with a unique approach to navigating music, and we need to get it to market in the most meaningful way."

Ultimately, that could mean an alliance with a digital music giant, though Marcus said the company could attempt an initial public offering within the next five years. With Apple being challenged in the digital music space by Napster, RealNetworks, AOL, MSN, Yahoo, Wal-Mart, and others, competitors could look to Savage Beast's technology to provide an edge, Forrester's Bernoff suggested.

''The nature of this digital music business is there are way too many companies chasing too few sales," Bernoff said. ''To survive, they need to differentiate themselves. And these recommendation technologies, to the extent that they work, could set them apart."

Robert Weisman can be reached at weisman@globe.com.