Title :
Using second-order hidden Markov model to improve speaker identification recognition performance under neutral condition
Author_Institution :
Electr./Electron. & Comput. Eng. Dept., Univ. of Sharjah, United Arab Emirates
Abstract :
In this paper, second-order hidden Markov model (HMM2) has been used and implemented to improve the recognition performance of text-dependent speaker identification systems under neutral talking condition. Our results show that HMM2 improves the recognition performance under neutral talking condition compared to the first-order hidden Markov model (HMM1). The recognition performance has been improved by 9%.
Keywords :
cepstral analysis; computational complexity; hidden Markov models; linear predictive coding; speaker recognition; speech coding; cepstral feature analysis; changing statistical characteristics; computational complexity; double stochastic process; linear predictive coding; neutral talking condition; observation vector; recognition performance; second-order hidden Markov model; state transition matrix; text-dependent speaker identification; Computational complexity; Decoding; Electronic mail; Hidden Markov models; Markov processes; Probability density function; Probability distribution; Speaker recognition; Speech analysis; Stochastic processes;
Conference_Titel :
Electronics, Circuits and Systems, 2003. ICECS 2003. Proceedings of the 2003 10th IEEE International Conference on
Print_ISBN :
0-7803-8163-7
DOI :
10.1109/ICECS.2003.1301992
