In this paper they preprocess a cent spectrum of an audio file. For comparison of audio files they create the Logarithmic Fluctuation Pattern(LFP)(p.3). For this I have to use a FFT on each frequency band of a block and then only take into account the amplitude modulations up to 600 bpm. (See screenshot)
I want to implement this LFP into Python but I have no idea what to look for.
Can I just use a simple Python
np.fft() on each block band?
What is meant with the amplitude modulation and the limitation of 600 bpm?
This is the only aspect of my work where I need this much insight and I tried to understand the amplitude modulation but it's seems to be pretty complicated.
Anyone can explain me what I have to do?