I am not sure if this already has been covered. I started to implement support for FFTW (Fastest Fourier Transform in the West, developed by MIT) and since i didn’t want to use the precompiled binaries i followed this link: mingw - Build FFTW lib with Visual Studio 2015 [added: steps for VS 2019] - Stack Overflow
It is not a 100% recipe for VS 2022 but 99.9%. It took me 10min to build the complete library (dynamic and static libtraries are possible, for the latter there is additional adaption of compiler switches necessary). Next step will be using this in JUCE.
Pls read the Licensing Model - using it free means GPL so you have to disclose all your source code)
1 Like
Interesting … I moved away from libFFTW to use the built-in JUCE fft methods instead, because of the license issues and also because its really quite cumbersome to maintain across multiple platforms - so I’m interested in what you did to overcome this. What is your motivation for doing this, out of curiosity?
Ok - I did spot a notice where it is pointed out that the Juce internal FFT has not the best performance. Afaik FFTW has areally good performace although I never tested that against the original Cooley-Tukey algorithm. Since it compiles under Linux and Windows this is ok for me.
It seems that Android is also possible (FFTW In Android NDK - Stack Overflow). Using dynamic linking means you only have to build the library once. The main issue is the GPL (if you do not pay for a license). My first 
project in JUCE is a GUI for a Hammond B3 Harware Emulator which shall be open source, so the GPL is no Problem.
I want to use the FFT for (almost) real-time analyzing of Organ Tone Generators.
1 Like
Makes sense - thanks for the explanation. Good luck with your project!