M. Neffe, T.V. Pham, F. Pernkopf, and G. Kubin (Austria)
Speaker verification, Gaussian mixture model - universal background model, voice activity detection, quantile filter ing, wavelet transform.
In this paper, a robust speaker verification system using improved voice activity detection has been designed for increasing safety of air traffic control. In addition to the usage of the aircraft identification tag to assign speaker turns on the shared communication channel to aircrafts, speaker verification is investigated as an add-on attribute to improve security level effectively for the air traffic con trol. The front-end processing unit is optimized to deal with small bandwidth restrictions and very short speaker turns. Two adaptive voice activity detection methods based on energy and wavelet parameters are developed and used as pre-processing in front-end unit. The verification task is accomplished by training background models and speaker dependent models. To enhance the robustness of the veri fication system, a cross verification unit is further applied. The designed system is tested with SPEECHDAT-AT and WSJ0 database to demonstrate its superior performance.
Important Links:
Go Back
