Roland Future Design Lab and Neutone have collaborated to launch Project LYDIA — a name which nods to the bridging of DIY and AI —which merges Neutone’s trainable AI software with Roland’s hardware expertise.
The current Project LYDIA mockup uses a Raspberry Pi 5 single-board computer which features “neural sampling” to learn the tonal qualities of any sound and then apply it in real time to another. This results in a huge range of sonic output, from musical to exotic and surprising.
The Pi runs a special version of Neutone’s Morpho alongside Roland’s latest ideas for a simple, tactile control panel. Audio input and output is managed via a USB Roland Rubix interface. Future versions aim to fully integrate everything into a self-contained device.
Roland has pioneered a DIY approach to music making — alongside its professional products, in 1981 it established the AMDEK (Analog Music Digital Electronic Kits) brand which offered a range of built-it-yourself kits from simple guitar effects to drum machines and mixers.
The accessibility of affordable and hackable hardware such as the Raspberry Pi and Arduino, together with applicable software, makes entry into experimentation for those with an enquiring mind and some practical skills even easier and more accessible than ever.
The term “neural sampling” means that users can capture sounds from the world around them, train a model on those sounds, and then use it in performance and recording. Any sound can be made ‘musical’ by morphing it in real time with an input carrier signal.
A primary goal of LYDIA is to make the technology into something that musicians could interact and play with. Working within the constraints of component sizes, heat generation, and more, Roland created a compact floor and tabletop design featuring simple, multipurpose rotary controls.
It’s worth noting that this is just the beginning, with prototype equipment which will be demonstrated at various events, and Roland inviting feedback from interested users.
More information can be found on the project page.
