The audio to be modulated is fed to VCA1 (upmiddle). The output of a second VCO (down) is fed into the second VCA´s signal input.

VCA2 (upright) is itself modulated by an LFO creating a pulsing spectral vibrato.

The entire spectrum is then used to modulate the first VCA.

Since we are using waveforms with high harmonic content (sawtooth) the resulting spectrum is a complex harmonic structure with a hundred sidebands.