I tried the program and it's really cool. I want to learn from it in order to make some FFT-based audio recognition.
My question is why the back-transform even works when the Frame class drops all the imaginary values and keeps only the real ones. Shouldn't that screw up all the phases in the audio?
I tried the program and it's really cool. I want to learn from it in order to make some FFT-based audio recognition.
My question is why the back-transform even works when the Frame class drops all the imaginary values and keeps only the real ones. Shouldn't that screw up all the phases in the audio?
Many greetings