Most of the paper is about media composition, neuron connection inducing, manipulation of neuron to promote binding to wires. There's some talk about neuron and neuron surrogate architectures for creating chemical artificial intelligence. However, the last figure and some of the issues they ran into seem relevant. The last sentence goes on to describing several methods they tried for removing/reducing the oscillations (Changing the power source to DC, adding a faraday cage around the biochip, reducing ions in the media). All method showed minimal improvement at best.
Also I've been thinking of the weird noise. What if it is just a spiral modulation on the frequency (Or oscillations that can't be removed) creating the nauseating effect and the weird, synthesized speech sounds.