TY - JOUR
AU - Nishikawa, Hiroaki
AB - Speech is one of the typical feature parameters and contains semantic information as well as singular information. So we can use semantic information for speech recognition, and singular information for identity verification. In order to realize secure and user-friendly human interface, both characteristics of speech should be utilized through simple procedures. In our paper, an identity verification method using speech is proposed. The proposed method utilizes CELP parameters which is used in speech coding scheme for mobile communication systems such as PDC in Japan, and can verify him/her only with the coded information. Therefore, following merits are given; (1) a speaker verification function can be easily realized in mobile terminals or networks by adding a little functions. (2) CELP parameters contain characteristics of articulation, so speaker can be verified whatever he/she speaks. We apply the latter merit to a challenge-response type verification named text-indicated speaker verification method which is realized by a text verification function as well as a speaker verification function, and both functions utilize only CELP parameters. In the proposed method, a system indicates texts that a user should speak at a verification process. So strength of security for disguise is increased. The reliability of the proposed method will be discussed with simulation results in this paper.
TI - Text-indicated speaker verification method using PSI-CELP parameters
JF - Proceedings of SPIE
DO - 10.1117/12.344667
DA - 1999-04-09
UR - https://www.deepdyve.com/lp/spie/text-indicated-speaker-verification-method-using-psi-celp-parameters-4Rg7d0nD3s
SP - 184
EP - 193
VL - 3657
IS - 1
DP - DeepDyve
ER -