In the clip, you see (and hear) a guy saying "bah bah bah" over and over. Afterward, he changes his tune to "fah fah fah" ... or so your eyes would have you believe. In reality, the audio never changed, only the picture did. That is, the voice is still saying "bah," but since it's now dubbed over a picture of the same guy pronouncing "fah," your brain actually changes what you're hearing so that it doesn't conflict with what you're seeing. If you close your eyes or look away, "fah" automatically goes back to being "bah."
Your brain also gave the "fah" version a tan, for unknown reasons.
This illusion is called the McGurk effect, and the creepiest part is that, even knowing know full well what's going on, you can't get your ears to hear the correct sound. Scientists who have been studying this shit for decades are still fooled.