If you can explain a bit more, I am interested in this. In addition, is it possible to copy the spectrum of other audio file with high precision with these fft libraries? Something like match EQ but super crazy!

 

First, "EngineersFilter" does about what your screenshot shows. It's a 32bit Windows VST but I'm sure you can handle that.
http://www.rs-met.com/freebies.html

Yes it's possible, it's called spectrum matching and there's already a number of plugins available that do this: Apply the spectral shape of one audio signal to another.
If you don't worry diving into a bit of programming, you can do a lot more, only limited by your imagination.
Matlab is another option that has been mentioned, but I don't have any experience with it.

My experience with such spectral experiments is that the crazier you go, the more you usually get audible artefacts that sound a bit like a super-badly compressed, unfiltered mp3 file. Your idea could be more exciting that the results you'll likely get but it's always possible to explore new territory if you have enough perseverance.

 

They're simple shapes actually, but the artefacts around the edges are a result of finite digital generation or display artefacts. So what you generate is not what you see. C https://en.wikipedia.org/wiki/Window_function#Windowing

 

Last edited: Jun 21, 2019

Are we moving into FFT yet? It's lovely there.

 

My best guess is we are in the middle of a convolution to FFT transition. stay tuned.

 

Sorry to say that but mathematically, it's nothing else than doing the same with a different algorithm.
You can do the convolution of your source audio with the recording of an impulse response.
You could also multiply the FFT of the source audio with the FFT of the impulse response.
Same result.
Which to decide for depends on your CPU and your means to optimise the calculation speed of the algorithm.

 

Nothing to be sorry about. I just point out that convolution and FFT are not the same, though efficient convolution it's is done using FFT.

You can do convolution with raw mathematics and no FFT. You can use the FFT and mathematically it's simple to perform a convolution but there are steps involved, including back and forth FFT/iFFT, multiplication/division in the frequency domain...

Zero-latency (relative to the sound card audio buffer) convolution is achieved using first raw convolution without FFT and the rest using FFT, block by block. As the IR is constant the IR FFT used in the convolution process is only performed once. And so on.

 

Last edited: Jun 24, 2019