An analysis of speaker recognition using bagging CAN2 and pole distribution of speech signals
ICONIP'10 Proceedings of the 17th international conference on Neural information processing: theory and algorithms - Volume Part I
Naive bayesian multistep speaker recognition using competitive associative nets
ICONIP'11 Proceedings of the 18th international conference on Neural Information Processing - Volume Part I
Improving bagging performance through multi-algorithm ensembles
PAKDD'11 Proceedings of the 15th international conference on New Frontiers in Applied Data Mining
Hi-index | 0.00 |
A method for speaker recognition which uses feature vectors of pole distribution derived from the piecewise linear predictive coefficients obtained by the bagging CAN2 (competitive associative net 2) is presented. The CAN2 is a neural net for learning efficient piecewise linear approximation of nonlinear function, and the bagging CAN2 has been shown to have a stable performance in reproduction and recognition of vowel signals. After training the bagging CAN2 with the speech signal of a speaker, the present method obtains a number of poles of piecewise linear predictive coefficients which are expected to reflect the shape and the scale of the speaker's vocal tract. Then, the pole distribution is used as the feature vector for the speaker recognition. The effectiveness is examined and validated with real speech data.