In order to analyze on phonetic characteristics of different speaker

put forward a method for speech discrimination based on bispectrum analysis. Sampling data of speech

corresponding processing

and carry out bispectrum analysis

comparing bispectrum of different speech

because of speech that contains different factors

characteristics of bispectrum is also different

the difference of result is obvious. Experimental results show that the apply bispectrum in speech analysis

can realize speech recognition of different speaker. It can greatly improve the accuracy of speech recognition

and it has a significant effect on processing of speech singles. Bispectrum is an effective method for speech recognition

and it has good application prospect.