This paper introduces Sirens, neural networks with periodic activation functions for implicit neural representations. The approach enables modeling of complex signals and their derivatives, solving partial differential equations like Eikonal, Poisson, and wave equations, with applications across images, video, sound, and wavefields.