This effect is a great example of the brain's effort to make sense out of conflicting sensory input. In this case, visual over-rides auditory. Other examples of this general (but fundamental) function include seeing faces in random images (or Jesus in burnt toast, depending on your predispositions…) or hearing voices or music in random noise.
This phenomenon also calls into serious question the general validity of personal experiences or testimony. In other words, the fact that you indisputably, undoubtedly heard "FAH" does not mean that "FAH" was spoken.