Temporal processing of auditory data plays a crucial role in our proposed model of tinnitus development through stochastic resonance (SR). The model assumes a physiological mechanism optimizing auditory information transmission (as quantified by autocorrelation (AC) analysis) into the brain by adding the optimal amount of neuronal noise to otherwise subthreshold signals. We hypothesize that this takes place at the second synapse of the auditory pathway in the dorsal cochlear nucleus (DCN). We propose that after hearing loss, this neuronal noise is increased in the affected frequency-band to improve hearing thresholds at the cost of upward propagation of this added noise, which finally may be perceived as tinnitus. We already showed the improvement of hearing thresholds in a large population of patients. Until now, we did not investigate the differences in hearing thresholds based on the biological constraints of early auditory temporal processing (phase locking) that is only possible up to frequencies of 5 kHz. In this report, we grouped our patient database (N=47986) according to tinnitus pitch (TP) of below (TP<5kHz) or above (TP>5kHz) the 5 kHz limit or having no tinnitus (NT) and compared their mean audiograms. We found that TP<5kHz patients showed significantly better hearing thresholds than all other patient groups independent of age. No improvement was seen for TP>5kHz patients who even showed worse thresholds than NT patients for high frequencies. These results are further evidence for our SR model of tinnitus development and the existence of AC analysis at the level of the DCN.