Holiday notice: Cycling '74 will be closed Monday, October 9th. Manual authorization, support cases, and manual order processing will be delayed.

Jul 15 2013 | 2:16 pm

Hey, so I was looking at the tutorial #25 in MSP, and I had several questions about using fft. When the sample analyzed is no longer a sinosuid or periodic, it seems that you have to window it to eliminate erroneous signals, but would the artifacts created by the FFT object and the windowing somehow tamper and mishandle the original data? I know you can use pfft~ also, but how would that affect the original data? Also, it suggests splitting data of samples that are longer than 1024 into different sections, but I am hoping to directly perform FFT on at least 10000 samples at a time, so how would the larger amount of samples affect analysis? I also don't understand why 512 samples is so important (well in the tutorial) and while I know the concept of fundamental frequency, how is 512 samples the fundamental frequency in this case? And in order for FFT to process it, does it have to be a signal first? would I use "*~" or "cycle~" or both? And if the number of samples is not a power of 2, which is needed for the algorithm of FFT, how will that affect how things run?