Research on Method of Speaker Recognition Based on Subtractive Clustering and Improved FCM

摘要

A speaker recognition method based on subtractive clustering and improved FCM clustering is introduced in this paper. MFCC and their difference MFCC are extracted from speech signal, secondly, subtractive clustering is utilized to initialize cluster centers, then improved FCM clustering is utilized to revise cluster centers and lastly code book is formed. In recognition, each unrecognized speech signal is clustered to the code books through FCM. The results of simulation experiments reveal that it has a higher recognition rate than that of single improved FCM and better robustness with simpler computation.