Disclosed are a microphone array-based sound signal processing method, apparatus and device. The method comprises: selecting, from a microphone array, a target microphone combination which is used for receiving a sound signal of a target sound source, the target microphone combination comprising a first microphone and at least one second microphone; obtaining target compensation information corresponding to the target microphone combination, the target compensation information comprising a signal compensation parameter of each second microphone with respect to the first microphone; according to the target compensation information, performing signal compensation processing on a second sound signal received by means of the second microphone; and according to the second sound signal subjected to the signal compensation processing and a first sound signal received by means of the first microphone, obtaining a target sound signal and outputting same.