Bob Smith (left) and James D. (JJ) Johnston (right) discuss FFT windowing at the January 2018 PNW Section meeting.
The PNW Section met in January 2018 to discuss windowing in FFTs. AES Fellow James D. (JJ) Johnston reviewed the theory of FFT windowing while PNW Section Technical Advisor Bob Smith showed practical usage. About 60 attendees came, 26 of which were AES members., and the meeting was held at the campus of Digipen Institute of Technology in Redmond WA.
Probably even neophyte audio DAW users have been taught how to find spectral frequency components in audio signals using some form of Fast Fourier Transform (FFT) in their software, but many have no idea how to choose from the FFT windows offered. Rather than a comprehensive primer of everything Fourier, JJ gave a simplified overview of the need for windowing in FFTs - that of the signal being analyzed being circular/continuous/repeating, and the effects of discontinuities. He spoke of how windows handle discontinuities, in effect low-pass filtering. Frequency precision is reduced, but now frequencies can be treated equally. He also spoke about rectangular, triangular, Blackman, Kaiser, Nuttall, and the Hann windows, and their characteristics. He generally likes the Hann window.
After a refreshment break, door prizes were handed out:
-Guitar straps, courtesy Ed Gruse - won by Doug Hogue and Rick Rodriguez
-Kensington trackball, courtesy Dave Lucavish/Eagle Computers, won by Jon Kasprick
-Swag bag and cable ties from NAMM, courtesy Dr. Mike Matesky/Opus 4 Studios - won by Ed Gruse
-Amprobe DMM, courtesy Rick Rodriguez/Fluke - won by Pete Dodds
-Pomona Volt Light, courtesy Rick Rodriguez - won by Koi Nil
-Fluke Hat Light, courtesy Rick Rodriguez - won by Thomas Cushman
Finally, PNW Section Technical Contributor Bob Smith used HpW Works Signal Analyzer software to show real-world FFT window results. He thought maybe 6 out of dozens of FFT windows are really useful for audio. Inevitably, every window has pluses and minuses. Bob showed how some windows will show a sharper frequency peak, and some will have different low amplitude resolution. As usual, there is no substitute for knowing a little about what you are trying to see in the first place as you try different windows.
