NSynth Super

NSynth Super is an experimental hardware instrument and software ecosystem developed by Google Creative Lab and the Magenta team. It utilizes the NSynth (Neural Synthesizer) algorithm, a deep learning model that uses a WaveNet-style autoencoder to learn the core characteristics of over 100,000 recorded sounds. Unlike traditional synthesizers that layer or filter audio, NSynth Super performs mathematical interpolation within the latent space of sounds to create entirely new timbres that possess the acoustic qualities of both source instruments simultaneously. Architecturally, the system is designed to run on a Raspberry Pi integrated with a custom PCB and a 4x4 touch interface, allowing users to navigate a grid of four distinct instrument types. As we move into 2026, NSynth Super remains a foundational reference for neural audio synthesis, offering musicians and sound designers a tactile way to explore the 'mathematical middle' of instruments—such as a 'flute-marimba' or a 'cello-glass' hybrid. The project is fully open-sourced, providing the community with the hardware schematics, firmware, and pre-trained neural models necessary to build and extend the technology.