One of the pieces of my rhythm stage setup is a breath controller, which measures how much I blow. It's a good fit for anything where I want to have some flexible imprecise control, but one thing I've ended up using it for a lot is just a pulse:

Here's it by itself:

Here's it live, layered with everything else I'm doing:

I'm still interested in having breath control as an option, for cases where I want more compelx patterns, but a basic pulse seems like something I should be able to automate. My system already knows the current tempo from how I'm playing the drums, so I could program it to simulate my breath pattern.

My first attempts at this sounded all wrong: it turns out I didn't actually know what pattern I was making. It was just something where I'd played around with blowing different ways until it sounded the way I wanted it. So I collected some data, tracking breath pressure as a function of how far through the beat I was:

I could do some kind of averaging to make nicer charts, but the overall picture is clear enough: I increase it linearly to a peak near the end of the beat (68% in reel time, 82% in jig time), and then decrease it linearly to a trough near the beginning of the next beat (27% in reel time, 41% in jig time). I programmed that, and now I can have the same pulse with my mouth free to do something else:

While I'm happy with how the result sounds, I'm pretty confused by the curves. The pulse sounds like it's on the upbeat, so I would have expected a peak at 50% (reel time), and I'd have expected the pictured curve to sound quite late? It seems like we identify the beat with the center of the trajectory (48% in real time, 62% in jig time)?