I don't think that is true, or even possible. It does not generate the sound you are about to hear. That would require prescience. It generates the sound that you have just heard, on the assumption that the next sound will be the same. That's why it works well for fairly continuous background noise.