This note presents a new Unknown Input Observer (UIO) with adjustable performances for linear systems. It focuses on state estimation in the presence of zeros with slow dynamics. It is known that if the system has stable invariant zeros and satisfies the decoupling condition then a UIO exists. However, the invariant zeros constrain the pole placement and consequently the state estimation error convergence rate, especially, when the zeros dynamics is slow. In the proposed UIO, the exact decoupling condition is relaxed by the introduction of the asymptotic decoupling notion which allows to act on the invariant zeros and then enhance the convergence rate of the state estimation error. Numerical examples are provided to illustrate the performances of the UIO.