Google's NSynth AI Creates Completely New Sounds

By: Joey Haar - May 16, 2017
References: magenta.tensorflow.org & digitaltrends
In the world of music, it's rare to hear something that's truly original -- and when those sounds do grace a track, it's likely a great find -- but a new Google project called NSynth is aiming to automate such novel noise. The program uses a neural network to combine the sounds of several different instruments and create completely unique, synthesized sounds.

To be clear, while NSynth does make use of sounds from multiple instruments at once, it doesn't literally compile those instruments' sounds on top of one another. Rather, the neural network gleans the mathematical properties of each instruments' sound waves, combining the noises on a far deeper level than if two people simply played a guitar and an electric keyboard out of the same amplifier.