Mapping 1D sound -> 3D volume in cylindrical coordinates:
- time - z
- intensity - r
- pitch class - phi
It is like the output of HarmonEye with time mapped to one axis instead of an animation.
It could be made quite precise by utilizing TF reassignment.
Also only the detected tones (without harmonics or percussions) could be displayed. Or just the percussions.
A similar but independent idea had also Peter Neubäcker, the founder of Celemony.