A synthesizer is a musical instrument that creates sounds using oscillators, filters, and samples; but it needs detailed parameters to adjust and tune them. Traditionally, this is done manually by analysing different attributes of real instruments, or by an expert who “handcrafts” the waveform. NSynth is a synthesizer that uses Artificial Intelligence to achieve this. A Deep Neural Network is trained with thousands of sample sounds to extract some features (such as Bright, Dark, Multiphonic, Percussive …) to which the parameters are associated. These parameters can be interpolated to generate a sound that averages these features, and not just the waveforms.

NSynth (Software)

NSynth (Launch demo)

NSynth Super (Hardware version)

Github repository (build instructions)