Domo arigato, Mr. Roboto

Found poem.   Reported in All Things Considered on December 22, 2009.

 

What was billed

as the first intercontinental musical interaction

between humans

and robots took place the weekend of Dec. 17.

It involved humans in Japan

using an application called ZoozBeat on their iPhones

and a robot named Shimon in Atlanta.

According to its makers,

 unlike other robots that can play music,

Shimon is perceptual.

The robot can listen

to what is played, analyze it

 and then improvise.

And it has been taught to improvise like some jazz masters.

Gil Weinberg of Georgia Tech's music technology

 program recently spoke to NPR's Robert Siegel from Japan,

 where he witnessed the historic interaction.

 Weinberg says the result is music meant to inspire people — not an effort

 to turn our music-making over to robots.

"The whole idea is to use computer algorithms

 to create music in ways that humans will

 never create," Weinberg says. "Our motto is, 'Listen like a human,

 but improvise like a machine.' "

Weinberg programmed Shimon

 to play like Thelonious Monk.

 He says that, though he and his team

were trying to teach the robot to play like a machine,

 they first had to teach it how a human plays.

 To do that, they used statistics

 and analysis of Monk's improvisation.

Once they had a statistical model

of the pianist, they could program the robot

 to improvise in that model.

Weinberg says the robot

 won't play everything exactly like the bebop pianist —

or any other jazz master —

would, though he says,

"It probably will keep the nature

 and the character of [the musician's] style."

"It's difficult to predict exactly

 what they would do in every single moment in time,"

 he says.

 "But our algorithm pretty much

 looks at the past several notes that it plays

 and, based on that, it sees what is the probability

 of the next note to be, based on all of this analysis

 of a large corpus of transcribed improvisation."

Some musicians are harder to program than others.

Weinberg says Ornette Coleman

would require a much larger body

 of transcribed work than Monk did.

"In a sense, it kind of reduces music

 to numbers and statistics," Weinberg says.

 Given enough tweaking to the algorithms

 that the program uses, he says

 he thinks they'll be able to create

 something "very similar to the jazz master."

But Weinberg says

 he doesn't think the robot

should try to play just like a human.

"In all the emotional

 and expressive energy,

 I don't think a robot can capture [it],"

Weinberg says.

 Maybe someday a computer program could,

 but at least right now, Weinberg says,

"I don't think we have the math for that.

We have some math to get the notes

and the rhythm and the scales.

Whether this can capture

the genius of Thelonious Monk,

I hope not.

But maybe."