Speaker recognition algorithm based on Gammatone filter bank
Mao Zheng-chon · Computer Engineering and Applications Journal · 2015
From the perspective of human auditory characteristics, a Gammatone filter bank model based on the auditory system in human cochlea is presented. This paper proposes a new Gammatone frequency cepstrum coefficients extraction algorithm, and in the Gaussian mixture model recognition system for the simulation experiment. The experimental results show that using the new algorithm to extract Gammatone frequency cepstrum coefficients, applied in the speaker recognition system, the system recognition rate and robustness are significantly improved.