Yes, you can, but you have to use a DSP and code frequency-domain filters like Fast Fourier Transformation that gives you the spectrum of the frequencies instead of the waveform of a sound. You can modify this spectrum (usually it is a buffer of 256 or 512 float entries) and then do a Inverse Fast Fourier Transformation (the inverse form of the previous function) to switch from the spectrum to the waveform.
However this approach is perfect for the results, but a bit heavy and eat a lot of CPU resources.
You may also use a cascade of IIR or FIR filters, each one working on a particular frequency.
You can find some good code for filters at http://www.musicdsp.org
I hope I was clear enough. Bye!
Please login first to submit.