I’ve experimented with audio DSP, first on the TMS320C40 and ADSP2181 in the late nineties. And then forgot about it for some time, because those were very limited being 16 bit DSP’s. Better DSP’s (like SHARC) were too complex to pour onto my own boards. I had fun with a SHARC devkit for some weeks, but the required board design was too complex to design my own. I picked up PIC programming, and doing audio again when the DSPIC came out, 16bit DSP in a microcontroller, building some algorithm blocks for sound synthesis again for some projects. Very limited and implementing DSP efficiently on those is a pain, but so cute.
Making it modularly configurable with a visual editor was a dream, but the projects I did with those did not have the room to build such a framework.
The ARM Cortex-M4 caught my interest, since it has the speed and architecture to support audio-DSP, and the microcontroller integration.
The first MIDI-controlled FM polysynth demo was quickly written at MusicHackDay Amsterdam in 2012, and the vision of visual sound algorithm editor came into my scope again. The original project page is currently unavailable, the musichackday team is working to restore old content on their new wiki.
iMinds, the independent research institute founded by the Flemish government to stimulate ICT innovation, has a call for projects for collaboration projects between arts and research, called “ART&D”. I thought this could be a good opportunity to develop this vision, and field test it with application-scenarios of different artists. I took initiative (as a researcher at EDM) to make a proposal with several artists.
The ART&D project named “Axo” started in spring 2013, and ended fall 2013. It was a rush to get to this point in 5 months of development time, dealing with hardware design, firmware and software all dependent on each other. It was a great drive to have a small user base very early in the development. And pleasure to see their joy and get their feedback.
A project getting shelved after finishing is a sad, so I kept on developing, and lauched a crowdfunding campaign with at talk at 31C3 (December 2014), reaching over 190% funding. Perks shipped starting August, with enthusiastic response. A community of users has grown, and Axoloti Core is now available.