As reported by Wired, the NSynth (Natural Synthesiser) is the product of artificial intelligence research at Google Brain, the company’s AI division. The program uses machine learning techniques to synthesise sounds from an original sample, resulting in a garbled, but altogether new sound.
Designed to aid the creative process, the NSynth generates sound at the level of individual samples from instruments such as organs, flutes and basses. Check out the examples below via Wired.
If you’re interested in the technicalities of the NSynth, have a look at the Google Magenta explanation here.