Music and fractal

Both music and fractals have multi-level structures. This makes me think of two possible ways converting fractals into sound or music:

Given a fractal on a 2d plane, we can view each vertical slice of pixels as a waveform or spectrum and scan over the fractal horizontally to produce a playable audio. It may require the fractal to be a real-valued map (instead of binary).

The waveform option feels more reasonable because it more likely to produce non-uniform characteristics on the sound, such as varying intensities across frequencies. However, it will be tricky to connect each waveform across time.

Another possibility is reading a fractal as symbolic music. By sampling through the fractal in a quantized grid, we can get a piano roll. But will that sound good? Or, we can view that as some kind of feature vectors over time?