Hi!

can we have audio spectrum chop outputting real and imaginary values?

Why?

For example I tried to patch an autocorrelation for BPM detection. Autocorrelation using FFT works more or less like this: You take the FFT of a time signal, you multiply this with the complex conjugate of itself and do the inverse FFT.

Now in TD, the closest I can get to complex number FFT is mag and phase output. So I need to do a polar to cartesian conversion, do my stuff, inverse convert, so cartesian to polar and then do the inverse FFT. Thats a source of errors, but it’s OK. But in order to output magnitude and phase, the audio spectrum CHOP also converts the internal complex numbers from cartesian to polar, and for the inverse FFT vice versa. So in total there are 4 conversions, which are all not needed in this case…

I imagine implementing this feature would not be hard since it means just not doing something the spectrum CHOP already does…

thanks and all the best