Magenta, a research team at Google, have built a machine-learning algorithm for creating new sounds, NSynth, as well as a controller for combining those sounds, NSynth Super.
Google have unveiled the NSynth Super, an open-source controller for creating sounds with the machine-learning Nsynth.
Artificial intelligence, machine learning and music production: Google’s algorithm for combining the timbre of different instruments can be controlled with the NSynth Super.
Aesthetically, the device doesn’t look dissimilar to the Korg Kaoss Pad, working in a familiar X/Y configuration to create new sounds from four source instruments.
In the video below, the device is demoed by Hector Plimmer. Sixteen source sounds, across fifteen different pitches, were fed into the NSynth algorithm, which then precomputed 100,000 new sounds.
The expansive new sound set was then loaded into the prototype device, with dials used to select different instruments and the touchscreen used to create unique blends of the tonal colours. Pitches can be played by any MIDI controller or DAW.
NSynth Super is an open-source project. For more information, check out the website.