Voice signal processing method, apparatus and server

Provided are a voice signal processing method, apparatus and server, relating to the technical field of communication. The method comprises: acquire an original voice signal from a plurality of channels, the original voice signals being digital voice signals (101); filter the original voice signal of each channel to obtain a first voice signal of each channel, the frequency of said first voice signal belonging to a pre-set frequency range (102); acquire from the first voice signal of each channel the loudness of the signal of each segment within the first voice signal (103); on the basis of a combination of the loudness of the signal of each segment within the first voice signal and the loudness of the signal of the same segment in the plurality of channels, acquire a first weight of the signal of each segment in the first voice signal (104); on the basis of the first weight of the signal of each segment in the first voice signal in the plurality of channels and the first signal of the plurality of channels, obtain a processed voice signal (105). The present voice signal processing method processes the signal by means of signal weight, improving the degree of voice recognition.