All the hits, all at once!

August 13, 2003 3:45 PM

Statistical analysis killed the radio star.Eigenradio analyzes the frequency content of 20+ stations at once, and mashes it, via math I don't understand, into music that is sometimes eerily beautiful, sometimes cryptically funky, and well, sometimes sounds like an Autechre CD stuck in a blender. Who says media amalgamation is a bad thing?
posted by arto (33 comments total)

I listened to this all afternoon. totally mesmerizing.
posted by jann at 4:11 PM on August 13, 2003

essence of clear channel. i listened for a while, then i had an odd urge to buy shit at walmart and contribute to a republican campaign.
posted by quonsar at 4:16 PM on August 13, 2003

jann and quonsar: I agree with both of you. (Thanks, arto)
posted by carter at 4:21 PM on August 13, 2003

look! the robot is singing!
posted by mcsweetie at 4:21 PM on August 13, 2003

But is it art?
posted by mmcg at 4:28 PM on August 13, 2003

I have the strangest urge to go out and buy a ThunderCougarFalconBird from FordNestleAOLNetscapeInstantMessengerExplorerDisneyPancakeLandWorld
posted by WolfDaddy at 4:41 PM on August 13, 2003

Thanks, arto. Now I have Statistical Analysis Killed The Radio Star stuck in my head. As I'm sure 98% of the rest of you do.
posted by mathis23 at 4:44 PM on August 13, 2003

Also by Brian Whitman: Ten Thousand Statistically Grammar-Average Fake Band Names.

His homepage gives the academic angle:
I'm interested in how language and music interact. We have shown good results in learning a description of audio from correlating music with internet opinion ("community metadata"), enabling a 'query-by-description' front end for music retrieval as well as insights into what certain words "mean" when used to describe music. I also work on "Long-Distance Song Effects," the path of a song's popularity and buzz over its short career, and ways to capture and predict this trajectory.

These problems tie together in the field of music acquisition: how to teach a computer about music from the ground up using unsupervised machine learning techinques and real-time audio analysis. (There certainly is no lack of data.) If a computer can teach itself enough about music on its own, we can hope for better results in retrieval and recommendation, as well more lifelike synthesis and coding.
posted by eddydamascene at 4:53 PM on August 13, 2003

Cool. Definitely has its moments.

Now if only there were some way to make math understandable to an English major. I'd love to understand what's happening.
posted by scarabic at 5:02 PM on August 13, 2003

1. get rid of the flanging
2. add some bass.
3. profit!
posted by signal at 5:12 PM on August 13, 2003

As they worked, Case gradually became aware of the music that pulsed constantly throughout the cluster. It was called dub, a sensuous mosaic cooked from vast libraries of digitalized pop; it was worship, Molly said, and a sense of community.
posted by syscom at 5:24 PM on August 13, 2003

That "What is is" chart looks an awful lot like something from DARPA.

Any connection?
posted by Etaoin Shrdlu at 5:34 PM on August 13, 2003

I can't seem to get it to work on my system. Darned system!
posted by dejah420 at 5:43 PM on August 13, 2003

Needs more cowbell.
posted by Acetylene at 5:48 PM on August 13, 2003

Take me to your leader.
posted by camworld at 6:10 PM on August 13, 2003

just wait until clearchannel tries to sue their punkass under the latest ThunderDMCAFalconPatriot frippery...

but in the meantime, it makes me feel like I'm at moma.
posted by dorian at 6:27 PM on August 13, 2003

Eigenradio sounds like Network Auralization for Gnutella pumped through a vocoder. Or John Oswald's Plexure (Sample, .au format).

Trippy.
posted by LimePi at 6:42 PM on August 13, 2003

Syscom, that is the most appropriate possible quote!
posted by josh at 6:50 PM on August 13, 2003

!!viva akufen!!
posted by Satapher at 8:11 PM on August 13, 2003

*head explodes*
Too much input!
posted by dg at 8:14 PM on August 13, 2003

Wow! It's actually pretty good. I wouldn't pay for a cd of it, but it's pretty fun to listen to knowing that it's synthetic, and it never repeats.

ISTR that John Cage wrote some pieces that called for perfomers with radios to switch to different frequencies at predetermined moments in the performance.
posted by electro at 8:34 PM on August 13, 2003

Yes, that sound composite is maddening and yet compelling at the same time. Fascinating. You know, I've always liked the sound of squeaky toys, even though they drive other people crazy. Squeek squeek squeek all day long, I love it! This sound reminds me of that same sensation -- annoying and yet thrilling. It kind of scares me. I wonder what parts of my subconscious it would tap into if I listened to it long enough. Thanks, Arto!
posted by PigAlien at 8:36 PM on August 13, 2003

everybody's got one ... everybody's got one ... everybody's got one ...

... take my purse.
If ever thou wilt thrive, bury my body;
And give the letters which thou find'st about me
To Edmund Earl of Gloucester; seek him out
Upon the British party: O, untimely death!

I know thee well: a serviceable villain;
As duteous to the vices of thy mistress
As badness would desire.

What, is he dead?

Sit you down, father; rest you ...
posted by yhbc at 9:05 PM on August 13, 2003

Just when I start questioning the point of this exercise, some remnants of blues harp waft in on top of a strong electro-acoustic beat, and it all makes sense.

Too bad it goes back to sounding like crap five seconds later. Perhaps the eigenvectors are improperly weighted. Or something.
posted by Galvatron at 9:44 PM on August 13, 2003

ISTR that John Cage wrote some pieces that called for perfomers with radios to switch to different frequencies at predetermined moments in the performance.

Yeah, it kind of reminds me of how, as a kid, I used to try and "play" that theremin-like whine between stations on the AM dial. Sounded kind of eerie, what with the actual stations on AM leaning towards either country music or hellfire-and-brimstone preachers at the time.

If it reminds me of anything out of the Gibson canon, though, it's not so much the dub favoured by the space rastas of "Neuromancer"--I envision that as being closer to the '70s spirit of King Tubby or Lee "Scratch" Perry--but the music played in Cognitive Dissidents, the bar from "Virtual Light". Don't have the quote handy, though...
posted by arto at 11:16 PM on August 13, 2003

This is amazing. Now if I could get my clock radio on the LAN here, I'd have wonderful antimusic to fall asleep to.
posted by sigma7 at 12:46 AM on August 14, 2003

Pretty cool. It sort of reminds me of the audio in the opening titles of the film Contact before it regresses into silence of space.

Anyone have some acid?
posted by Down10 at 2:13 AM on August 14, 2003

Hey syscom, I'm just re-reading Neuromancer these days.

How did you know that?

[Looks around nervously, then gets sucked into the Eigenradio, analyzed for any interesting frequencies, and spit back out into cyberspace.]
posted by Outlawyr at 4:24 AM on August 14, 2003

Sounds uncannily like shortwave radio interference. Probably created in a similar way?
posted by asok at 6:02 AM on August 14, 2003

Whitman records as Blitter on Hrvatski's reckankomplex label. Just a couple of EPs and contributions to various comps so far, but a full-length is on the way.
posted by Dean King at 7:38 AM on August 14, 2003

Is this a joke? I listened for about 6 or 7 minutes, convinced it was very cool, when suddenly I heard Dr Dre's 'Bad Intentions' come in mid-chorus. It played on, unmodulated, with no interference, to the very end of the song.

Now tell me that for 3 minutes, that song's every note and beat possessed the most entropy of anything else on the radio...
posted by yellowcandy at 4:50 PM on August 14, 2003

What, is he dead? Sit you down, father; rest you ...

No, Paul is not dead. However, John and George (and Linda) are.

Great link, arto. I listened for 10 minutes (until the top of my head started to come off), and will return often.
posted by LeLiLo at 9:40 PM on August 14, 2003

Now tell me that for 3 minutes, that song's every note and beat possessed the most entropy of anything else on the radio...

Funny, I listened to it last night and heard Sean Paul's "Get Busy" (or at least recognizable, longish snatches thereof) in amongst the static twice within a half-hour. Which probably means something, but I couldn't tell you what... (Damn you, Vibe 98.5. Daaaamn youuu!)
posted by arto at 11:13 PM on August 14, 2003