'Musical Genome' Helps Find New Favorites The company Savage Beast's "music genome project" catalogues songs to the smallest detail in a bid to more accurately link people to music that they might like. Scott Simon talks with Tim Westergren, the founder of Savage Beast technologies. An interactive product will be available soon.
NPR logo

'Musical Genome' Helps Find New Favorites

  • Download
  • <iframe src="https://www.npr.org/player/embed/4661167/4661182" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript
'Musical Genome' Helps Find New Favorites

'Musical Genome' Helps Find New Favorites

  • Download
  • <iframe src="https://www.npr.org/player/embed/4661167/4661182" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript


This is WEEKEND EDITION from NPR News. I'm Scott Simon.

You've heard of the Human Genome Project. Well, now get ready for the Music Genome Project. A company in Oakland called Savage Beast is cataloguing songs down to the smallest of details and determining how they relate and complement one another. Savage Beast uses a computer program that evaluates each song by 400 distinct musical attributes, not just, `Well, it's got a nice beat. It's easy to dance to.' So, let's say for a music lover who enjoys "Proud Mary"--and that's certainly stretching the term--the program can link to other music that shares some of "Proud Mary's" 400 distinct qualities, songs that they venture a person would probably also like. So joining us from our studios in San Francisco is the founder of Savage Beast, Tim Westergren.

Thanks very much for being with us.

Mr. TIM WESTERGREN (Founder, Savage Beast Technologies): My pleasure.

SIMON: So now how is this different than, say, ordering a Hank Williams CD on Amazon, and then the next time you log in Amazon says, `Well, you'd probably, in addition to "Hank Williams' Greatest Hits," you'd probably like his lesser hits and music by Ernest Tubbs'?

Mr. WESTERGREN: Yeah. This a very, very different approach. What we've done, actually, is using people, trained musicians, we've listened to now an enormous amount of music and analyzed each individual song based on existing musical properties; as you mentioned, close to 400 unique musical characteristics. So we're capturing everything from melody and harmony to rhythm, form, compositional qualities, lyrical content. It's really a detailed taxonomy of music, so essentially capturing all the details that give the piece of music its individual sound or style.

SIMON: I--we mentioned "Proud Mary" in the beginning. I mean, with all due regard to a big hit and a song that's important to millions of people around the globe even today, does "Proud Mary" even have 400 attributes?

Mr. WESTERGREN: Yes. Every song, every song--yes, it does. Yes, I mean, every song can be understood along these many, many dimensions. And what we're really trying to do is capture anything about a piece of music that might be a reason why somebody likes it.

SIMON: But let's listen to an example.


(Soundbite of "Long December")

Mr. ADAM DURITZ (Lead Singer, Counting Crows): (Singing) The smell of hospitals in winter and the feeling that it's all a lot of oysters with no pearls.

SIMON: Now this is "Long December" by the Counting Crows.

(Soundbite of "Long December")

Mr. DURITZ: (Singing) And all at once look across a crowded room to see...

SIMON: OK. So if you like "Long December" by the Counting Crows, what might come up?

Mr. WESTERGREN: Well, one recommendation would be a song called "Tonight" by an artist named Ben Arthur.

SIMON: All right. Let's take another short listen, and then we're going to cross-fade into Ben Arthur's "Tonight."

(Soundbite of "Long December")

Mr. DURITZ: (Singing) Drove up to Hillside Manor sometime after 2 AM and talked to a little while about the year...

(Soundbite of "Tonight")

Mr. BEN ARTHUR (Singer): (Singing) A pile of shadows and broken glass, a shimmer borne on the air...

SIMON: It does sound a little bit similar. What am I hearing?

Mr. WESTERGREN: So there are many elements that make these what we call `musical neighbors.' There's instrumentation, there's melodic contour, there's similar harmony, they're both very diatonic. But there's one particular element that I think is quite intriguing. They both avoid resting too much on the tonic, which is the root chord. And the effect of that to a listener is to create something of a meandering quality. And that's not something that maybe as a layperson you'd be aware of, but you'd be--or you would not know what to call it--but you'd be aware of that in this song. And that's actually a characteristic of both of those artists.

SIMON: How do people find you and take advantage of this?

Mr. WESTERGREN: Well, in a couple of months this product will be fully available in a very interesting form--unfortunately, I can't discuss in too much detail right now--where our listeners will be able to have a conversation with the genome.

SIMON: For those of us who don't know quite what that means, what does that mean?

Mr. WESTERGREN: There will be a product that will be available for someone to log into and actually interact with this, create--and put songs they like, listen to music, have a full sort of genome conversation on the Web.

SIMON: Another example of how this might work. Let's start with an old Cros--Well, I guess they're all old, aren't they?--Crosby, Stills & Nash song.

(Soundbite of "Helplessly Hoping")

CROSBY, STILLS & Nash: (Singing) Helplessly hoping her harlequin hovers nearby awaiting a word.

SIMON: OK, Mr. Westergren, rock my world.

Mr. WESTERGREN: So one thing this--that CSN is well-known for is their very interesting use of harmony. In this, using the Music Genome Project, you could input that song and look for recommendations. And if you did that in a database, and you said, you know, `I like this song, and I particularly like the sound of the vocals,' we come up with a recommendation from a band called Rogueway(ph), and they use very similar group vocal dynamics.

(Soundbite of unidentified song by Rogueway)

ROGUEWAY: (Singing) Please recant and repent. Uh-uh-uh-uh-ooh.

SIMON: Yes! I mean, they sound as if they were influenced by Crosby, Stills, Nash.

Mr. WESTERGREN: Which they undoubtedly were.

SIMON: Mr. Westergren, we mocked "Proud Mary" a little bit and that's just not right. Could you do "Proud Mary" for us?

Mr. WESTERGREN: Absolutely.

SIMON: Thank you.

Mr. WESTERGREN: I would be happy to. I assume you want the CCR version of that.

SIMON: Yeah, Ike and Tina is what we had in mind.

Mr. WESTERGREN: Oh, Ike and Tina.

SIMON: Yeah.

Mr. WESTERGREN: I'm going to try and type that in here. OK, so here's what you get with that.


Mr. WESTERGREN: "Proud Mary" brings up "10th Avenue Freeze Out," by Bruce Springsteen. "Had Me a Real Good Time," by The Faces; let's see. "Gloria," by Patti Smith.

SIMON: Oh, yeah.

Mr. WESTERGREN: "Sweet Maxine," by the Doobie Brothers.

SIMON: This is fun, yes.

Mr. WESTERGREN: "Movin' Out," by Billy Joel.

SIMON: Yeah. Well, Mr. Westergren, this has been a lot of fun. Nice talking to you.

Mr. WESTERGREN: My pleasure. Thanks.

SIMON: Tim Westergren is the founder of Savage Beast Technologies, and you can find out how to take advantage of that service by going to our Web site, npr.org. And if you like our Web site, let me tell you about a few more.

(Soundbite of "Proud Mary")

Ms. TINA TURNER (Singer): (Singing) All right now, Mary!

Background Singers: (Singing) Rolling...

Ms. TURNER: (Singing) Yeah!

Background Singers: (Singing) ...rolling...

Ms. TURNER: (Singing) Uh-huh!

Background Singers: (Singing) ...rolling on the river.

Ms. TURNER: (Singing) Give me one more time!

Background Singers: (Singing) Rolling...

Ms. TURNER: (Singing) Yeah!

Background Singers: (Singing) ...rolling...

Ms. TURNER: (Singing) Whoo!

Background Singers: (Singing) ...we're rolling on the river.

Copyright © 2005 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.

NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.