What I'm trying to do is see if a particular frequency component becomes dominant (and I don't really know what the dominant frequency is).
Therefore, I figured that I can get the top two peaks of an FFT, and then get the ratio of the power between them. If the ratio becomes too high, then there is a dominant frequency in the spectrum. However, I cannot use the FFT due to memory and real-time computation constraints, and I really only want the top two frequencies anyways, so it doesn't make sense for me to compute the entire spectrum.
I've tried using Pisarenko's method, but I couldn't get the resolution I needed. The FFT peaks will probably be at least 25 Hz apart and my sampling rate is 12KHz.
Thanks!