Summary: A phase coupled oscillator neural network with amplitude modification and a phase shift is studied as a model of sequence-associative memory. The dynamical properties of the pattern recalling process are very important. With regard to the equilibrium state of the system, there have been many studies based on the replica method. Also, an exact solution without any approximations has been proposed for sequence retrieval processes by \textit{A. Düring} et al. [J. Phys. A, Math. Gen. 31, 8607-8621 (1998; Zbl 0962.82056)]. In that work, only a self-averaging property is assumed, and a set of self-consistent equations is derived for a stationary state. In this paper, we extend their theory and derive the dynamical equations governing the macroscopic order parameters for general maps. These maps include both amplitude modifications and phase shifts under a synchronous updating rule. As in the case of the Ising spin model, it is found that the crosstalk noise is distributed in a Gaussian manner for the sequence-associative \(XY\) spin model. In order to verify this, we consider four different maps. The results of the Monte Carlo simulations for these maps agree well with the theoretical results and support the validity of the theory.