The processing works like this: the raw audio data is run through digital signal processing using a Wiener Filter, where you can start to hear some information. This is then further cleared up through a neural model, giving the researchers clear audio.