This paper proposes an innovative method of feature extraction of Spectrogram using Pulse Coupled Neural Network(PCNN) in speaker recognition for the first time.Using the invariable characteristics of PCNN time series and entropy series of Spectrogram,people can extract the feature of speaker's speech and recognize the speakers rapidly and effectively.The validity of this method is well verified by authors' experiments.It is significant for theoretical research and application of PCNN to bring PCNN into spe...