This is a visualization tool for music temperament. The complex shapes that appear with sounds on the black canvas are called Lissajous curve. This curve represents frequency ratios between notes that compose the chord. Classical temperaments (like just-intonation temperament) uses integer ratios for many intervals. This means that the Lissajous curves doesn't move in many chords for the classical temperaments, unlike the modern temperament (equal temperament), which uses an irrational number (2^(1/12)) for the interval ratios.
In this tool, the single sound wave of each note is alternately accumulated into Lissajous x and y sound wave streams in the order you play. The shape represents most recent (1024 / sampling_rate (= 44100 in most case)) [sec] sound wave.
This work is inspired by this video originated(?) from Justonic Tuning Inc products. I don't have the product, so I guessed how it works and emulated it. (It's unbelievable that the instructions of the video seems to be using the real sounds from the microphone. How is it possible to extract frequencies with such accuracy? Someone tell me.)
Source codes (GitHub) Any suggestions and issues are welcome.