Really fast audio in iOS and Mac OS X using Audio Units is hard, and will leave you scarred and bloody. What used to take days can now be done with just a few lines of code.

Novocaine *audioManager = [Novocaine audioManager];
[audioManager setInputBlock:^(float *newAudio, UInt32 numSamples, UInt32 numChannels) {
	// Now you're getting audio from the microphone every 20 milliseconds or so. How's that for easy?
	// Audio comes in interleaved, so,
	// if numChannels = 2, newAudio[0] is channel 1, newAudio[1] is channel 2, newAudio[2] is channel 1, etc.
}];
[audioManager play];

Novocaine *audioManager = [Novocaine audioManager];
[audioManager setOutputBlock:^(float *audioToPlay, UInt32 numSamples, UInt32 numChannels) {
	// All you have to do is put your audio into "audioToPlay".
}];
[audioManager play];

Yep. Novocaine is result of three years of work on the audio engine of Octave, Fourier and oScope, a powerful suite of audio analysis apps. Please do check them out!

The RingBuffer class is written in C++ to make things extra zippy, so the classes that use it will have to be Objective-C++. Change all the files that use RingBuffer from MyClass.m to MyClass.mm.

Inside of ViewController.mm are a bunch of tiny little examples I wrote. Uncomment one and see how it sounds.
Do note, however, for examples involving play-through, that you should be using headphones. Having the
mic and speaker close to each other will produce some gnarly feedback.

If you want to get down and dirty, if you want to get brave and get close to the hardware, I can only point you to the places where I learned how to do this stuff. Chris Adamson and Michael Tyson are two giants in the field of iOS audio, and they each wrote indispensable blog posts (this is Chris's, this is Michael's). Also, Chris Adamson now has a whole gosh-darned BOOK on Core Audio. I would have done unspeakable things to get my hands on this when I was first starting.