A dual speech/speaker recognition using GMM in speaker identification and a HMM in keyword speech recognition

摘要

In this paper, a speaker recognition voice based on GMM system is presented. We test the system using several databases recorded in several sessions in order to repair the huge effects that the speech variability with time has in the recognition rate system. Several experiments have been made in order to achieve the best configuration in the system set up, and in the selection of the amount and distribution of training speech. This is an important point to take into account in a real world system in which users train the system once and the models generated in the training process are not updated for strategic reasons. The dualities provide and additional security requirements in the applications in which this security level is necessary. In this sense the system provides in a real implementation approach an error around 5%, that is a very interesting rate in a real environment.