Google recently created a prototype for artificial intelligence hardware in the form of a musical instrument (a synthesizer), but you won’t be seeing this product for sale. They designed it, but you have to build it if you want one. Thankfully, Google is providing the plans.
Revealed on Tuesday, this instrument makes unusual sounds based on real sounds. Apparently, some people at Google wanted to create a hardware version of a software synthesizer they created last year.
This software was an AI system that came out last spring called NSynth, where N is for neural. It’s “trained on hundreds of thousands of short recordings of single musical notes played on different musical instruments,” CNBC reported. This data allows the software to create sounds of different notes that are longer, shorter, or blends of multiple instruments.
Etch A Sketch for Sound
NSynth is a foundational technology from Magenta, a Google Brain AI research group looking into making art and music with AI. AI Duet and sketch-rnn are other Magenta projects. An open-source version of the NSynth software was released that allowed a person to blend two instruments. This blended sound could then be played using the keys on a computer keyboard.
The hardware synth prototype, called NSynth Super, has multiple physical knobs and displays that are more attuned to the needs of live performers. It goes further with controls to adjust qualities like attack, decay, sustain and release. Operators are also not limited to two instruments– NSynth Super lets you combine up to four instruments at once. Basically, it offers more choices and a higher level of precision than just the old NSynth software.
“We wanted to make an Etch A Sketch for sound,” Joao Wilbert of Google’s London Creative Lab, said. Wilbert was one of the people who worked on the hardware.
While Etch A Sketch is basic with two knobs, the similarities exist. There are four knobs on the corners of the prototype to choose instruments. You then drag your finger across the display to blend the sounds. As logic would have it, taking your finger to any of the four corners would produce the sound of that instrument, whereas anywhere in between would give you a combination of sounds. The AI creates the strangest sounds closer to the middle.
With a computer hooked up, a backdrop can be applied to which you could follow the same process. By connecting a piano keyboard and creating the perfect filter, you could play out your own music using that sound.
This technology has an appeal even to non-musicians. It’s as simple as amusement and providing an ability to explore creation.
Why wouldn’t Google try to capitalize off of this product by producing and selling it? The prototype is a research project rather than a source of revenue like the Pixel phone or Home speaker. Google’s essentially showing off what is possible with AI software they made– creating just to create.
While Google’s NSynth Super has beautiful metal housing, the average consumer is more likely to get one made of wood or plastic. It can also run with the simple Raspberry Pi miniature computer. The purpose of this simplicity is accessibility. Because it can be built more cheaply, more people can afford it.
AI can play a helpful role in the wholly human creative process.
Toyota engineers, meanwhile, looked into basketball in their free time. They created Cue, a 6 foot, 3 inch tall humanoid robot that uses AI to calculate and shoot perfect free throws.
As humans, we try to create better intelligence trying to rival our own. Instead of fearing it, we might look at applications like NSynth Super and Cue to see how development of this technology should be celebrated as an expansion of our capabilities and our world.
Hardware and software for Google’s prototype are online on GitHub.