The invention discloses a hearing compensation method for high-frequency hearing loss patients. The method comprises the following steps: voice signals are converted into input signals; noise reduction and end point detection are performed for input signals so as to judge and distinguish the sound section and the noise section; the signal spectrum and the noise spectrum are divided into six frequency domains; according to the proportion relation of the spectrums of the different frequency domains of the signal section and the noise section, standardization is realized through being replaced by 1/2 power; and frequency-reduction voice output signal is adaptively derived from input signal through slow-releasing factor and is converted into the frequency-reduction voice. The hearing compensation method can better maintain the tone information of the Chinese speech, improves the understanding degree of hearing loss patients to the Chinese speech, and averagely improves the recognition rate by 10-15 percent for the Chinese speech and environment sound.