Google wants musicians, machines to create \"new kind of art\"

While numerous studies may suggest that machine-learning algorithms will put human workers out of work soon, it appears like singer-songwriters won’t be among the latter. Google Brain, as a part of its Magenta initiative, is currently working on tools that pair artists with deep-learning tools in order to develop novel artwork together.

What’s Magenta?

Magenta is a Google Brain project to ask and answer questions, “Can we use machine learning to create compelling art and music? If so, how? If not, why not?” The work done under Magenta is done in TensorFlow and the company regularly releases models and tools in open source. These are further accompanied by demos, tutorial blogs and technical papers that allow readers to follow their progress.

Google established Magenta in order to achieve two goals: First, to advance the state-of-the-art in music, video, image and text generation. Second, to build a community of artists, coders and machine-learning researchers. In order to facilitate these goals, the core Magenta team is building open-source infrastructure around TensorFlow for making art and music. This already includes tools for working with data formats like MIDI, and is expanding to platforms that help artists connect with machine learning models.

Research scientist at Magenta Douglas Eck shared his experience at MIT Technology Review’s EMTech Digital Conference by saying that it was fun trying to create art with AI that responds and learns from humans. “I don’t think that machines themselves just making art for the art’s sake is as interesting as you might think,” he said. “The question to ask is can machines help us make a new kind of art?” he added.

However, the team currently faces a critical challenge wherein they are now trying to develop better human interfaces for the technology. Initially, the researchers started with the equivalent of a command-line prompt, but researchers at Magenta say that they want to get closer to the “naturalness” of a particular musical instrument. The team hopes that the project attracts talented musicians and coders in order to continue enhancing tools and apply it in newer ways.