two sine tones, summed and pushed through a nonlinear stage — a speaker driven hard, an eardrum, a fuzz circuit, all shrug the same way — and pitches appear that neither oscillator is playing. Giuseppe Tartini heard the loudest one, |fA−fB|, on solo violin double-stops in 1714 and called it his "terzo suono," the third sound. its quieter twin fA+fB rings alongside it. this isn't synthesized separately — it's genuinely in the signal, the same intermodulation math your ear does to any two loud tones.
0 ghost = clean linear mix, no ghosts possible · turn it up to fold the sum through a soft nonlinear curve · "stop" fully kills both oscillators and closes the audio graph