When using adaptive active noise control (ANC), the performances of the noise reduction, speed of convergence and stability is highly dependent on the adaptation algorithm. While some of the most popular algorithms are the NFxLMS (Normalized) and derivations of VSS (Variable step size), the gra-dient descent optimization could benefit from the recent algorithms developed for the neural network research, knowing that the more advanced algorithms are computationally intensive. Meanwhile, with modern hardware, it is now possible to apply some of these algorithms in real-time for the adaptation of the active noise control algorithm. This paper compares the performances of some recent algorithms in terms of noise attenuation, convergence speed, robustness and sensitivity to parameters selection.