Our story
- 2011
DrumPi began with Paolo, who simply couldn't afford a proper electronic kit. He'd been a software engineer for years when he had the idea that a smartphone could serve as the sound engine for a drum module.
The computational power was there, but the latency wasn't. Even iOS devices with Core Audio couldn't get below 6.5 ms once buffer size, the driver's ring buffer and USB polling were accounted for. The result was decent, but no better than any reasonable computer could manage.
- 2015
Paolo abandoned the smartphone and started building prototypes on a Raspberry Pi 3, working at a much lower level. Latency dropped from 6.5 ms to 2.5 ms — but the code sat so low that developing it further became painful. It was a Ferrari with no seats and no steering wheel.
- 2019
After a few years away sharpening his modern C++, Paolo came back to the project with help from his friend Max. Paolo rewrote the audio engine; Max took on UI and UX. They swapped the Linux kernel for preemptive-rt, profiled execution times, and isolated one core of the quad-core CPU to run nothing but the real-time audio thread.
In roughly ten months the audio engine had become a DAW in its own right: a 64-channel mixer, effects, MIDI, multi-layered and multi-velocity samples, a transient shaper, and multitrack audio and MIDI recording.
The core was done — but how did it feel to play? Paolo isn't a professional drummer and couldn't judge it. So they brought in Valerio and spent months tuning the response and feel. Making Valerio happy was no easy feat; he had a keen ear for authenticity. Given the results, it was worth every bit of the effort.
