How Google is making music with artificial intelligence

2 downloads 0 Views 854KB Size Report
Aug 8, 2017 - software developed in part by Google's. Magenta GOOGLE. How Google is making music with arti4cial intelligence. By Matthew Hutson Aug.
Home

News

Journals

Topics

Careers

Search





Log in

SHARE

News from Science has introduced metered access. Full access to all news content is included in AAAS membership.

 1K

  12



A musician improvises alongside A.I. Duet, software developed in part by Google’s Magenta GOOGLE

How Google is making music with arti4cial intelligence By Matthew Hutson

Aug. 8, 2017 , 3:40 PM

Got a tip?

How to contact the news team

Related Jobs ScienceInsider New rule could force EPA to ignore major human health studies BY WARREN CORNWALL

Can computers be creative? That’s a question bordering on the philosophical, but artiIcial intelligence (AI) can certainly make music and artwork that people Ind pleasing. Last year, Google launched Magenta, a research project aimed at pushing the limits of what AI can do in the arts. Science spoke with Douglas Eck, the team’s lead in San Francisco, California, about the past, present, and future of creative AI. This interview has been edited for brevity and clarity. Q: How does Magenta compose music?

APR. 25, 2018

With €1.5 billion for artiQcial intelligence research, Europe pins hopes on ethics BY TANIA RABESANDRATANA

APR. 25, 2018

U.S. EPA says it will deQne wood as a ‘carbon-neutral’ fuel, reigniting debate BY SCIENCE NEWS STAFF APR. 24, 2018

A: Learning is the key. We’re not spending any effort on classical AI approaches, which build intelligence

Salk puts cancer scientist Inder Verma on leave after harassment allegations, announces investigation

using rules. We’ve tried lots of different machine-learning techniques,

BY MEREDITH

including recurrent neural networks, convolutional neural networks,

WADMAN

variational methods, adversarial training methods, and reinforcement learning. Explaining all of those

Trump’s EPA wants to stamp out ‘secret science.’ Internal emails show it is harder than expected

buzzwords is too much for a short answer. What I can say is that they’re

BY SCOTT WALDMANN, E&E NEWS, NIINA HEIKKINEN, E&E NEWS

all different techniques for learning by example to generate something new. Q: What examples does Magenta learn from?

APR. 21, 2018

APR. 20, 2018

More Science ScienceInsider

Sifter

A: We trained the NSynth algorithm, which uses neural networks to synthesize new sounds, on notes generated by different instruments. The SketchRNN algorithm was trained

Gene therapy scores success at treating thalassemia

on millions of drawings from our Quick, Draw! game. Our most recent music algorithm, Performance RNN was trained on classical piano performances captured on a modern

Scientists accidentally found this octopus nursery deep in the PaciQc Ocean

player piano [listen below]. I'd like musicians to be able to easily train models on their own musical creations, then have fun with the

Amateur mathematician cracks decades-old math problem

APR. 19, 2018

APR. 18, 2018

APR. 18, 2018

resulting music, further improving it.

00:00 00:00 Q: How has computer composition changed over the years? A: Currently the focus is on algorithms which learn by example, i.e., machine

Global trove of rare earth metals found in Japan’s deepsea mud APR. 13, 2018

Humans may have developed bat foreheads to communicate with eyebrows APR. 10, 2018

learning, instead of using hard-coded rules. I also think there’s been increased focus on using computers as assistants for human creativity rather than as a replacement technology, such as our work and Sony’s “Daddy’s Car” [a computercomposed song inspired by The Beatles and ]eshed out by a human producer]. Q: Do the results of computergenerated music ever surprise you? A: Yeah. All the time. I was really surprised at how expressive the short compositions were from Ian Simon and Sageev Oore’s recent Performance RNN algorithm. Because they trained on real performances captured in MIDI on Disklavier pianos, their model was able to generate sequences with realistic timing and dynamics. Q: What else is Magenta doing? A: We did a summer internship around joke telling, but we didn’t generate any funny jokes. We’re also working on image generation and drawing generation [see example below]. In the future, I’d like to look more at areas related to design. Can we provide tools for architects or web page creators?

More Sifter

Magenta software can learn artistic styles from human paintings and apply them to new images. FRED BERTSCH

Q: How do you respond to art that you know comes from a computer? A: When I was on the computer science faculty at University of Montreal [in Canada], I heard some computer music by a music faculty member, Jean Piché. He’d written a program that could generate music somewhat like that of the jazz pianist Keith Jarrett. It wasn’t nearly as engaging as the real Keith Jarrett! But I still really enjoyed it, because programming the algorithm is itself a creative act. I think knowing Jean and attributing this cool program to him made me much more responsive than I would have been otherwise. Q: If abilities once thought to be uniquely human can be aped by an algorithm, should we think differently about them? A: I think differently about chess now that machines can play it well. But I don’t see that chess-playing computers have devalued the game. People still love to play! And computers have become great tools for learning chess. Furthermore, I think it’s interesting to compare and contrast how chess masters approach the game versus how computers solve the problem—visualization and experience versus brute-force search, for example. Q: How might people and machines collaborate to be more creative?

A: I think it’s an iterative process. Every new technology that made a difference in art took some time to Igure out. I love to think of Magenta like an electric guitar. Rickenbacker and Gibson electriIed guitars with the purpose of being loud enough to compete with other instruments onstage. Jimi Hendrix and Joni Mitchell and Marc Ribot and St. Vincent and a thousand other guitarists who pushed the envelope on how this instrument can be played were all using the instrument the wrong way, some said—retuning, distorting, bending strings, playing upside-down, using effects pedals, etc. No matter how fast machine learning advances in terms of generative models, artists will work faster to push the boundaries of what’s possible there, too. Posted in: Engineering, People & Events, Technology doi:10.1126/science.aan7216

Matthew Hutson Matthew Hutson is a freelance science journalist in New York City.

 

Email Matthew Twitter

More from News

Einstein’s ‘spooky

With €1.5 billion for

Solar cells that work

‘spooky action at a distance’ spotted in objects almost big enough to see

billion for artiQcial intelligenc e research, Europe pins hopes on ethics

that work in low light could charge devices indoors

Science 20 April 2018

Get Our Newsletters GENETICS

Omen in the blood

Vol 360, Issue 6386

ANTHROPO LOGY

Ancient DNA untangles South Asian roots

ANTHROPO LOGY

Cannabis, opium use part of ancient Near Eastern cultures

SCIENTIFIC COMMUNIT Y

Subscribe Today

Proposal to rescue postdocs from limbo draws darts

Receive a year subscription to Science plus access to exclusive AAAS member resources, opportunities, and beneIts.

SCIENCE AND POLICY

Plan for 2020 U.S. census is fatally Xawed, critics say

First Name Last Name Email Address

Enter your email address below to receive email announcements from Science. We will also send you a newsletter digest with the latest published articles. See full list

✓ ✓ ✓ ✓ ✓ ✓

Science Table of Contents Science Daily News Science News This Week Science Editor's Choice First Release NotiIcation Science Careers Job Seeker

Email address By providing your email address, you agree to send your email address to the publication. Information provided here is subject to Science's Privacy Policy.

ATMOSPHE RIC SCIENCE

Departmen t of State's air pollution sensors go global

Table of Contents

Subscribe Today

Sign up today

About us Journals Leadership Team

Advertise Advertising kits Custom publishing

For subscribers Site license info For members

International

Help

Chinese Japanese

Access & subscriptions Reprints & permissions

Stay Connected









Contact us Accessibility

members Work at AAAS

© 2018 American Association for the Advancement of Science. All rights Reserved. AAAS is a partner of HINARI, AGORA, OARE, CHORUS, CLOCKSS, CrossRef and COUNTER.

Terms of Service

Privacy Policy

Contact Us