Arpeggiator

An arpeggiator makes no sound. It sits between your hands and whatever instrument is mounted, takes the notes you are holding, and plays them one at a time on a grid. That distinction matters here more than usual: this module has no voice of its own, so what you hear is the preview engine's own synth voice being played very quickly by something other than your fingers.

The pattern is a function of the held set, not a recording of it. Add a note to a chord you are already holding and it joins the pattern at the next step rather than restarting it; lift one and the cycle shortens around the gap. The step grid runs free from the moment the first key goes down, so the rhythm never stutters when the harmony changes underneath it.

Random is seeded, which is a deliberate departure from the plugin this was ported from. There, each step called the system random number generator, so the same patch played a different line every time and no two renders of it matched. Here the choice is a hash of the step index: the line is still unpredictable to listen to, but it is identical on every playback, in an offline render, and in the test suite. Musically it costs nothing; it is what makes the mode reproducible.

This is a preview module. The scheduling is real, every note is committed to the AudioContext clock as an absolute time, so timer jitter cannot move it, but the voice it drives is the website's stand-in engine, not a compiled kernel, and no parity claim applies.