This is one of many videos from the artist Prabu Deva (aka Benny Lava) that has been posted on YouTube.
People have gone through and transliterated what they believe he is singing that is, the words at the bottom of the clip are not what are actually being sung, but what someone thinks it sounds like. Try listening to the video with your eyes closed.
Can you tell what he is singing? Now watch the video and read the captions. Watch the video a third time, again with your eyes closed. Can you tell what hes singing this time? How is this an example of top-down processing?