Maker here. mdrone is a drone instrument: you hold one sustained tone, layer a few voices, and let it evolve for minutes or hours.
With microtonal tuning the partials beat against each other and "phantom" melodies start to surface that you never actually played.
It ships several tuning systems: Pythagorean, several just-intonation sets, Kirnberger III, 31-TET, plus maqam-leaning scales like Yaman and Bayati, and a custom scale editor.
The same held interval is a different sound in each.
Tech: it's all client-side Web Audio + AudioWorklets, no backend, no account. The whole scene (voices, tuning, FX, two slow modulators) serializes into the URL, so a sound is just a link you can bookmark or share. There's a small adaptive-stability layer that eases off visuals/FX/voice count if it detects the audio thread struggling, to avoid dropouts on weaker devices.
Free and open source (AGPL): https://github.com/gdamdam/mdrone