Tutorials and Topics - Peabody Computer Music

Tutorial 25Analysis:Using the FFTThe triangular window is simple and quite effective.) In this way, the fft~ object isviewing each time slice through a triangular window which tapers its ends down, thusfiltering out many of the false frequencies that would be introduced by discontinuities.Overlapping triangular windows (envelopes) applied to a 100 Hz cosine waveTo accomplish this windowing and overlapping of time slices, we must perform twoFFTs, one of which is offset 256 samples later than the other. (Note that this part of thepatch will only work if your current MSP Signal Vector size is 256 or less, since fft~ canonly be offset by a multiple of the vector size.) The offset of an FFT can be given as a(third) typed-in argument to fft~, as is done for the fft~ object on the right. This results inoverlapping time slices.One FFT is taken 256 samples later than the otherThe windowing is achieved by multiplying the signal by a triangular waveform (stored inthe buffer~ object) which recurs at the same frequency as the FFT—once every 512samples. The window is offset by 1 /2 cycle (256 samples) for the second fft~.• Double-click on the buffer~ object to view its contents. Then close the buffer~ windowand double-click on the capture~ object that contains the FFT of the windowed signal.Scroll past the first block or two of numbers until you see the FFT analysis of thewindowed 1000 Hz tone.199