A recovery apparatus for solving a branch mis-prediction, and a method and a central processing unit thereof are provided. The wireless power receiver includes a receiving unit receiving the power from the wireless power transmitter using resonance, and a rectifying unit rectifying the power received therein from the receiving unit to supply the power to a load side. In this method, an emphasized vocal tract characteristic sp2 is generated to output by applying formant emphasis, using amplification ratios calculated based on estimated formants, directly to the vocal tract characteristic sp1 for instance. The audio output from each of the host computers may be mixed or switched for output to the speakers as desired.