Modeling the frequency dependency of different defining parameters, such as time- and length-scale plays an important role in the accuracy of predictions for the noise emanating from a turbulent flow, such as jet flow. Many research workers have considered this problem, usually using a mathematical model for the form of the frequency dependence which does not necessarily stem from the physics of the turbulence, but is instead chosen to fit measurements or to give a good acoustic prediction. In this paper the underlying physics of the turbulent flow is considered in the derivation of a suitable time-scale, and this is shown to have the merit that frequency dependence arises naturally. This time-scale is shown to be related to the turbulent energy transfer rate. The time scale is then used in the prediction of noise from an isothermal subsonic coplanar jet working at two flow velocities. Essential fluid dynamic information and turbulent parameters have been obtained using a modified k- ε method, and the jet noise directivity is studied using the MGBK method. The results clearly show an improved noise prediction after using the new time-scale in comparison with some more traditional definitions.