the-kandinsky-project
A system that reads one of Kandinsky's Bauhaus paintings and plays it as music, using a colour-to-sound logic built from his own theory in the books On the Spiritual in Art (1911) and Point and Line to Plane (1926). A machine-learning model extracts the painting's elements — points, lines, planes, their colours and positions. This reading is currently semi-automated. Each element is then turned into sound by rules drawn from Kandinsky's writing, where he claimed colour and form obey a law as exact as music.
Pitch is relational: no colour owns a note. The background sets the tonal home, and each element's pitch comes from how far its colour sits from that home in a perceptual colour space — close is consonant, opposite is tense — so the music's harmony is a direct readout of the painting's colour relationships. Height sets register, colour chooses the instrument, and a multi-colour plane sounds as a chord.
Kandinsky called his correspondences a science, then left the decisive choices — how long a point lasts, how loud a diagonal is — to feeling. Making his theory actually play means filling those gaps, by measurement where his text allows and by ear where it doesn't. A synchronized video highlights each element as its sound plays, so you can watch the painting being read. First in a series that hands progressively more of the reading to the machine, and eventually turns the same apparatus on other painters who wrote down their own correspondences.
