In a blog post, we describe the details of NSynth (Neural Audio Synthesis), a new approach to audio synthesis using neural networks.
There are three main ways to use NSynth:
-
An interactive AI Experiment made in collaboration with Google Creative Lab that lets you interpolate between pairs of instruments to create new sounds.
-
A MaxForLive Device that integrates into both Max MSP and Ableton Live. It allows you to explore the space of NSynth sounds through an intuitive grid interface. [DOWNLOAD]
-
Create your own audio samples with the pretrained network.
See also the Jupyter notebook for NSynth in the /jupyter-notebooks directory.