I'll toss this out here and let others run with it.. I don't have a lot of time now... but here's the idea:
A microphone picks up sound and converts it to an electrical
signal. If you view a signal on a plot it looks like a wavy line. For example,
http://blogs.mathworks.com/seth/2009/01/30/mathworks-conversations-and-the-fft/
The job of audio equipment is to transfer that signal from one form to another without distorting it. A tape recorder puts that signal in a permanent form on a tape. A CD burner puts it on a CD. A CD player or DAC turns it back into an electrical signal. An amplifier attempts to make it larger (more powerful) without distorting it. A headphone changes it back into sound, hopefully without distorting it too much.
But none of these devices functions perfectly. They all distort the signal in some way.
The frequency response
plot (*) is an attempt to characterize the way in which a device distorts the signal. See this FR plot of a microphone:
http://www.totalvenue.com.au/articles/microphones/microphones.html
It's based on this observation:
- play pure tones through a device and vary the frequency of the tones from very low to very high
- don't change the loudness of the tones
- in a perfect device the output tone will stay the same loudness
- but in an imperfect device, the output loudness will vary. Most obviously, at very low frequencies the output will drop to nothing. And at very high frequencies, too.
- Plot that variation and presto you have a FR plot
Now you might ask this: what good is characterizing the behavior with pure tones, seeing as music is made of complex signals?
Well in the 19th century a mathematician named Fourier made a remarkable discovery: if you know the FR of a device, you can predict its response to
any signal.
That's right---run some pure tones through it, and presto you've just described its behavior for all time.
But there's a huge caveat: the device must have a certain property called
linearity for this mathematical proof to hold. And no real device is linear. Nonetheless, most real devices are close enough that the FR plot is useful.
There's a lot more to say about how the FR corresponds to perception, but this is the basic theory.
(*) The wikipedia article describes FR as a range of a low number to a high number. That's a variation on the idea I'm describing here. I am describing the foundation of the theory so you have a deeper understanding of what those numbers mean.