This repository provides a Chirp-Z transform in Python.
The Chirp-Z transform provides a way to sample the frequency domain representation of a discrete signal. Rather than sampling the entire frequency domain (as the FFT does), it allows sampling of a sub-range, with a configurable frequency resolution.
The Chirp-Z transform can be implemented using three "normal" FFT invocations.
The Chirp-Z transform is an interesting alternative to using zero-padding to increase the spectral resolution of the frequency domain. It uses a lot less memory, needing only FFTs with size (m + n - 1) or greater. Our implementation selects the smallest power-of-two that is greater than or equal to (m + n - 1), as power-of-two FFTs are particularly efficient.
The ChirpZ transform also can be used to calculate an N-point FFT when only a power-of-two FFT is available. This provides a reasonably efficient way to calculate FFTs for any integer value N.