Chris Darwin was doing a research paper on speech perception and discovered something interesting. He distorted recordings of people speaking based on specific pattern matching properties of speech, and then played them back for people to see if they could understand what was said.

At first, it sounds like distorted tones and bleeps. But after you listen to the original version – your mind seems to be able to hear the speech clearly on the second playthrough. It turns out the brain’s prediction/learning knows better what to expect, and applies that without the listener being consciously aware.

If one listens to enough different SWS samples, some begin to gain the ability to understand the SWS versions without having to be exposed to the originals.

Go here for details on SWS as well as to the (free) Praat software that Darwin used to create the samples. [Hackaday]

A little longer examination: