I mentioned neural networks in the sense that Dr Shore models the brain parts responsible for tinnitus as a feedforward neural network, and uses this model to arrive at the correct timings. So to replicate the timings we'd need to come up with a similar model, which would require serious neural networks expertise.Neural networks? That would be quite interesting, but I'm unsure about how it would work with so few samples. Do you have a source?
But in this case we're talking about a literal network of living neurons, not neural networks in the machine learning sense (although that would surely be cool).
For example, see this study: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2566901/