In a recent publication, we show that the algorithm of Matching Pusuit is applicable to estimating arrivals within room impulse responses. The study of the time distribution of arrivals allows one to define a time beyond which arrivals are statistically distributed. We call that time the cross-over time. However, these estimates are extremely sensitive to the stopping criterion of Matching Pursuit - which is closely linked to the precision of the approximation. When the precision is too low, we arrive at an unrealistic number of arrivals. On the other hand, when the precision is too high, the model of room impulse responses gives poor estimates. The best stopping criterion can be determined based on a comparison of room acoustical indices or using listening tests. In this paper, we use both approaches in order to estimate arrivals and the cross-over time of experimental room impulse responses of a concert hall from their Matching Pursuit decomposition. We show how perceptually determining the stopping criterion leads to a perceptual estimation of the cross-over time. We are thus able to make comparisons with the theoretical formula of the mixing time proposed by Polack, which is related to the volume of the hall.