 Can a complicated signal like the sound of my voice really be built out of individual sinusoids? Let's try an experiment. Here's a sinusoid in the time domain. And here it is in the frequency domain. Doesn't sound like much though, does it? Adding an additional sinusoid at higher frequencies changes the signal, but it still doesn't sound like my voice. Let's add in some more sinusoids around the frequencies already playing. Now things are starting to sound interesting. Let's add some more sinusoids and some more. As I add more and more sinusoids, the signal begins to sound like my voice until Fourier transform. This is what the inverse Fourier transform does. Build a signal out of individual sinusoids.