Fractional Fourier transform (FrFT) is one-parametric family of unitary transformations . FrFT are usually interpreted as rotations in the time-frequency plane. In case FrFT becomes an ordinary Fourier transform ( ), in case we have inversion transform ( ), for FrFT corresponds to inverse Fourier Transform ( ), and for (or ) we have identity transform ( ). The family forms a one-parameter continuous unitary group with additive multiplication . FrFT corresponds to decomposition of a signal into linear combination of chirps (or signals in which the frequency increases (up-chirp) or decreases (down-chirp) with time). FrFTs have found wide application in various signal and image processing tasks, however the problem of improving their fast algorithms remains relevant. In this work we develop fast algorithms for discrete fractional and four-parametric Fourier transforms (FPFT) with minimal computational error. For that, we derive kernels for these transforms using a method of projective decomposition of the discrete Fourier transform operator (DFT). Analysis of computational complexity shows that the complexity of obtained algorithms is comparable to the fast Hartley transform (FHT, ).