We introduce an efficient time/amplitude warping tool, which was originally developed for registration-based waveform inversion. Time and amplitude warping functions are modeled with piecewise cubic polynomials and they are obtained by solving a nonconvex optimization problem iteratively in a multiscale fashion from low to high frequencies. In order to seed the frequency sweep at zero frequency, low frequency augmented signals are used for the optimization problem. Such low frequency augmented signals are generated via nonlinear transformations. We investigate the effect of different low frequency augmented signals on optimization. We also demonstrate time/amplitude warping of complicated seismic traces and applications of time warping to 4D data.