Under some circumstances (simple signal or hardware-conditioned signal),
simple routines can be surprisingly useful.

1) If you're just interested in the lowest frequency, and/or 
2) if the part of the signal you're interested in has the highest
amplitude, 
3) then you could look for the average peak-to-peak time in the sampled
signal, then invert to get frequency. Longer samples, better resolution.

In the case of large duty-cycle square waves, you could switch to
detecting the average time between leading edges.

There is a Mac shareware out there called "Guitar Tuner" that does a
pretty good job with just the old "lapel" type Mac mike. Playing with it
by varying the quality of note you play (or sing or bang) into it can
help you to understand some of the possible problems.

This is a large and interesting subject; modems (phone and radio) face
many of the same problems with audio. Historically, filtering has
probably been the most important part of solutions. (E.g., detecting
radio signals bounced off the moon.)

Questions: What built-in filtering does the Mac OS or hardware have? Is
Inside Macintosh the best place to learn more about it? How about
learning what samples look like in RAM and how to modify them?

-- Tony