Why FM detectors produce noise that rises with audio frequency (the triangular noise spectrum), so the high end of the audio band would come out noisiest. To fix this, the transmitter runs the speech through a pre-emphasis network, a simple high-pass shaping circuit (about 6 dB per octave, typically a 75 microsecond time constant) that lifts the higher audio frequencies before modulation. The receiver then applies the mirror-image de-emphasis, restoring flat response while pushing the high-frequency noise back down.
Watch out A de-emphasis network is the matching circuit at the receiving end, cutting the highs back down; it attenuates rather than boosts. The other two choices are not real circuit names.
PRE-emphasis comes first (transmit, boost); DE-emphasis comes after (receive, cut).