The system was trained on people listening to stories. It decoded words from the brain data recorded while people were watching a silent video. For some of the videos we used, we actually had these audio description tracks. So when we decoded words in those movies, we could then compare it to the audio description tracks and ask, like, was it similar statistically? And it was. It was actually decently correlated with what was in the audio description track. But surprisingly, it really does kind of describe events that are happening in the movie,. I was quite surprised by this, and I think it's pretty cool.