Credit: Unsplash
And they said AI couldn’t make beautiful music.
In today’s technologically-enhanced world, just knowing how to play a guitar isn’t quite enough (though it is cool). To really get the most out of an electric axe, you gotta know a thing or two about sound theory, and that includes the proper use of an effects pedal to get all those cool wah-wahs squeedly-dees. To properly use an effects pedal, it needs to be carefully programmed ahead of time, then manually activated with one’s foot at the proper part of a song. If you’re not much of a multitasker, perhaps you need a little assistant to handle the modulation while you focus on playing the actual guitar.
Programmer and guitarist Keith Bloemer developed his latest pet project with that precise goal in mind: a Raspberry Pi-powered AI guitar effects pedal. Bloemer created a neural network from scratch by feeding it information about sound modulation, as well as a variety of different effects and techniques. The result was NeuralPi, an AI effects pedal that can emulate all sorts of existing guitar effects, as well as create entirely new ones from scratch. It can distort, amplify, reverberate, and a theoretically endless amount of other kooky effects.
NeuralPi is a guitar pedal that uses neural networks to emulate real amps and pedals on a @Raspberry_Pi 4: https://t.co/9zWXZYBnII pic.twitter.com/I1tTPpuScl
— Edge Impulse (@EdgeImpulse) May 31, 2021
The really interesting part about NeuralPi is that assembling it didn’t even require any particular esoteric parts. Bloemer’s device cost him about $160, including a Raspberry Pi 4, a HiFiBerry ADC +DAC, and a HiFiBerry case. Oh, and a male/female RCA cable to plug it all in, of course. You can assemble the device yourself fairly easily, and Bloemer has even made the source code available on GitHub. You will also need to download Elk Audio OS, a special open-source operating system designed to be used with sound and music-based projects.
