AI systems for music are overwhelmingly digital – they exist on screens, and are often controlled with text prompting that makes them fundamentally distinct from the ways that humans have been making music for thousands of years, which are with physical and acoustic instruments. Not only that, generative AI music in its current form is usually soulless, hyper-commercial, and threatens the livelihoods and the craft of human musicians.
The Automatar is built to change that – it uses AI in all the ways that current AI systems do not – it keeps the human in control, it is fundamentally physical and acoustic, and is committed to craft over easy gratification and instantaneous output.
It is the first instrument known to the creator that uses an NVIDIA GPU to run AI models literally on-instrument – no WiFi, no API, no surveillance by mysterious companies in the cloud. This is fundamentally a tool – it will not change overnight or suddenly disappear like so many AI products.
It’s trained with small datasets that enable it to harmonize by playing itself electromagnetically, amplified, but fundamentally acoustic. It doesn’t try and copy human behavior – in fact, its use of electromagnetism ensures it plays in a way that no human can.
The Automatar uses AI to create performances that were impossible before.
