I've been doing some research into naturally-powered music makers recently, and came across this very cool paper/schematic for "Solar Sounders". Even if you aren't interested in building something like this, I find the author's explanation of how he turned certain acoustic sounds into a circuit to be really fascinating.

Instead of offering a universal palette of sounds, the circuits focus on specific acoustic sounds. In this work, the analysis of three “voices” begins the task of emulation. For a solar-modified sonic environment, I chose a bird, Tibetan monks and the trains I hear at night. Consider the sound-making apparatus of each — a throat for the monk and the bird, and a horn for the train. Imagine its interface, how to control it and how it makes sound, to conceive a model for its timbre. Then consider the triggering brain and what triggers it to sound — territorial concerns, meditation, emergency or other contingencies. In the emulating medium (circuits or code), work backward from the brain, ending with the physical sound becoming electronic.