A Brief History of Grackle

The week immediately proceeding Halloween 2024 is a blur to me, as I spent pretty much every day chatting with Naomi about Grackle and then staying up till sunrise trying to hammer out the software implementation details. Luckily, there are chat logs and some dubious commit messages to reference:

November 1st: Sound is working, envelopes are working, intervals are working, the different algorithms are working.

November 2nd: ActualFM input works, tested three different modes for the auxiliary output that eventually became mod out, fixed some issues regarding inverted CV ins, made envelope attack logarithmic and decay exponential. Also a commit saying “It’s like, pretty much feature complete????” (hilarious and no absolutely not)

November 3rd: Doubled the number of interval options on the sliders, drastically reduced mod amounts because it was way too easy to push Grackle into “noise town,” adjusted mod and interval CV ranges to match the slider ranges, various bug fixes and tweaks.

November 4th: Increased buffer size and reduced CV polling frequency to get rid of some troublesome noise/bleed on the CV inputs. This would come back to bite me later.

November 5th - 8th: Soooooo much testing and patching. Realized that having both sliders control the amplitude of each modulator in MUL mode was redundant due to the commutative property and we decided to have the first mod depth slider control a triangle wavefolder affecting the product of the modulators instead.

Past the 8th, Grackle largely resembled the current incarnation, so you might ask what we’ve been up to since then. In my experience, software feature development moves pretty quick (especially when you’re so excited that you’re burning the candle at both ends and a secret third end) but refinement and hardware revisions necessarily have to move at a slow and steady pace.

As such, the last seven months have involved a combination of gathering feedback, making hardware tweaks, ordering parts, planning, documenting, web development, video production, code-hardening and debugging, optimization and refinement, and sooooo much patching.

The most notable hardware improvements from that intervening time were adding envelope normalization to the depth inputs, tweaks to the LEDs, and getting the auto-calibration hardware in order (including the adding the shift button, which allowed us to introduce interval-unlocking and plan some potential extra features down the road). Software refinements included getting the CV polling rates well into audio range, 2x supersampling of the operators, refining the Serge/LPG-inspired envelope generator, and ironing out what happens when you send negative CV to the depth and interval inputs. Oh, and so much latency/code optimization, because getting the Daisy to run stable with a 2-sample-wide buffer does not provide much computational wiggle room.

Ever since that fateful Halloween, Naomi and I have been excitedly sending each other videos of all the cool sounds we’ve managed to wring out of Grackle. The video above is a best-of all the little snippets I’ve sent to Naomi over the last several months.

Grackle has been such an unexpected and incredible source of excitement and joy for myself and Naomi both. I know we were both struggling in mid-to-late 2024, and I think it would be fair to say that Grackle’s development really turned things around. Personally, it’s rekindled my love of modular synths and further solidified my ongoing obsession with phase modulation synthesis.

I’m so grateful to Naomi for giving me the privilege of helping bring this idea into reality, and I’m so pleased that the development of this module spurred us to found Entropic Loop together. She’s an incredible synthesist, designer, musician, writer, and all around person and I am so lucky to know her and have the unique privilege of working together with her creatively.