FFTs,
tes
ant
So you aren't talking about splicing them together then. You're talking about simply taking overlapping blocks from the data stream. That's so simple (in concept at least) that I don't think you'll find much literature devoted specifically to how to do it. You'll find mention of it in various ap notes for at least the box-type dynamic signal analyzers, and in the help text. It comes down to using overlap when you can: if your data stream is 100Ms/s and you are doing 1M point FFTs, you need quite a bit of processing power to keep up. "Keeping up" is commonly called "real-time processing" or something similar. So the bandwidth that supports real-time for some particular length FFT could be an important feature when you go to buy an instrument. Just how much overlap you'd like to have depends on how much you want to be sure that you don't miss anything, and what window function you're using. So again, since all of this is tied up so intimately with the whole concept of an FFT-based analyzer, I'd suggest an ap note about the fundamentals of such analyzers: sections
8 and 9 of Agilent ap note 243Cheers, Tom