Provided are a voiceprint recognition method based on a voiceprint library, and a master control node and a computing node. The voiceprint recognition method based on a voiceprint library, which method is applied to a master control node, comprises: receiving a 1:N voiceprint recognition request, wherein the 1:N voiceprint recognition request comprises voiceprint data to be subjected to recognition, and N is the capacity of a voiceprint library (S11); forwarding the 1:N voiceprint recognition request to at least two computing nodes, so that each of the computing nodes allocates, according to the 1:N voiceprint recognition request, voiceprint data stored in an embedded memory database of the computing node to at least two CPU cores of the computing node for parallel processing of the voiceprint data to be subjected to recognition, so as to obtain a voiceprint recognition result (S12); and receiving the voiceprint recognition result returned by the at least two computing nodes (S13). According to the method, the time complexity of voiceprint recognition can be reduced, and the processing performance of a voiceprint recognition system is thus improved.