FFT is actual very technical but typical in audio processing - see:

http://en.wikipedia.org/wiki/FFTMay be you can find a web-site explaining it in your language.

So the FFT analyses gives you the energy per defined frequency bands, e.g. an BASS_DATA_FFT1024 analyses results in an float array with 512 elements.

Each element represents the energy value of a certain frequency (see Nynquist - yes, this is audio theory again, but it helps you to understand ;-).

The index of the element in the array can be calculates as followed:

Example: If the stream is 44100Hz, then 16500Hz will be around bin 383 of a 1024 sample FFT (1024*16500/44100).

So if you are looking for frequencies within the raw PCM sample data you need to perform an FFT analyses, as the raw PCM sample data just doesn't contain it straight away.

The raw PCM data just describe the wave curve over time...