Mammals integrate multiple timing cues to differentiate natural sounds such as vocalization sequences. We are building Bayesian computational models that successfully use statistical correlations of timing cues in natural sounds to accurately predict perceptual behavioral decision-making. Our models predict improved sound duration judgment when when accumulating more information, supporting a principle where animals flexibly integrate and judge timing cues on multiple scales. Our measures of cortical neuron spike-rate responses to timing cues support the hypothesis that perceptual judgment of temporal sound sequences could be mediated flexibly by primary (A1) and higher level supra-rhinal auditory field (SRAF) cortices which integrate onset and duration timing cues on distinct log-log (power law) scales. Our second Bayesian computational model incorporates these neural spike-rate response properties and predicts behavioral decision-making and judgment and attributes a rank order decrease in accuracy between A1 and SRAF.