Two-stage decision for short utterance speaker identification in mobile telecommunication environment

The short utterance speaker recognition is necessary and convenient to ensure proper access to private information, personal transactions, and security of computer and communication networks. The performance of the traditional speaker recognition in mobile telecommunication environments, such as cellular, PDA and telephone environments, degrade seriously because of the coding effect in the telecommunication, which impair the speaker´s feature information. So some work needed to be done to improve the recognition performance in the mobile applications. In order to achieve better results in cellular and telephone environments, the paper proposes a two-stage decision for short utterance speaker identification. And the paper evaluated the method in our SRMC (speaker recognition for mobile communication) corpus.