Four voices, sixteen steps, one bar. The drums are synthesized from oscillators and filtered noise — there isn’t a sample on this page — and the tone voice takes its pitch from the step it sits on, so melodies come from where you put things, not what you choose. It starts with a polite default pattern. Clear it.

Kick
Snare
Hat
Tone

Built with the Web Audio API and nothing else: a lookahead scheduler so the timing holds when the tab breathes, envelopes as exponential ramps, and a noise buffer that gets band-passed into being a snare or high-passed into being a hat, the way drum machines have faked it since the seventies.

It plays sound only when you press play, and stops when you leave. That shouldn’t be worth stating, but here we are.