This is an interesting article on CD reproduction of high-frequency noise, especially the graph. I hadn't realized that the approximation was that far off.