Well, there isn't really a proper board for this, so I'll post it here... I'm sorry if this is the wrong section ;).
I've implemented FFT. I understand it. But my question is: how best to use it?
I'd like to input some audio samples and get the frequencies that are actually most present in this sound (or the part of the sound). I've tried to input a simple sine function and while it does report the proper frequency to be most present, the frequencies before and after it seem to be pretty 'present' as well. That is, the intensity of the proper frequency is highest, but those of a few hundred Hz more or less are quite high as well. Is there a proper way to distinguish properly between them?
Also, what is the best number of points (samples) to use?
And finally, what is the best way to input the data. Let's say I'm using 4096 points (samples) (I currently am). Should I input the first 4096 samples, then the next 4096? Or is there a better method, for instance: first input 4096, then the last 2048 of the last buffer and the next 2048 bytes, and so on? So, should the input overlap or not?
Thanks in advance,