Speaker Recognition for DSR
نویسندگان
چکیده
Due to the coexistence of different compression algorithms in the fixed and mobile telephone networks, it is impossible to predict which combination of coders and channels the speech has undergone before arriving to the server. To overcome the previous mentioned problem, the European Telecommunication Standards Institute (ETSI) has standardized a front-end for Distributed Speech Recognition (DSR). But once again, the distortion added due to feature compression in the front-end side increases the variance flooring effect that increases the identification error rate. The penalty incurred in reducing the bitrate is degradation in speaker recognition performance. In this paper we present a non traditional solution for the previous mentioned problems. To reduce the bitrate, speech signal is segmented at client and the most effective phonemes for speaker recognition are selected to be sent to the server. Speaker recognition is occurred at server. Applying this approach on YOHO corpus, we could achieve 0.05% identification error rate (ER) using an average segment of 20.4% of the testing utterance for recognition. This result outperforms previously published results on the speaker identification task from error rate (ER) point of view as well as the minimum speech segment required for speaker identification.
منابع مشابه
Speaker recognition and the ETSI Standard Distributed Speech Recognition Front-End
With the advent of Wireless Application Protocol (WAP) and 2.5/3G communication systems, the mobile device has become a window to the Internet. A natural interface to this mobile device is through speech. To address this need, a new European Telecommunications Standards Institute (ETSI) standard front-end has evolved for Distributed Speech Recognition (DSR). The goal of the ETSI DSR front-end i...
متن کاملAn Innovative Distributed Speech Recognition Platform for Portable, Personalized and Humanized Wireless Devices
In recent years, the rapid growth of wireless communications has undoubtedly increased the need for speech recognition techniques. In wireless environments, the portability of a computationally powerful device can be realized by distributing data/information and computation resources over wireless networks. Portability can then evolve through personalization and humanization to meet people’s ne...
متن کاملSpeaker Independent Speech Recognition Using Hidden Markov Models for Persian Isolated Words
متن کامل
Speaker Adaptation in Continuous Speech Recognition Using MLLR-Based MAP Estimation
A variety of methods are used for speaker adaptation in speech recognition. In some techniques, such as MAP estimation, only the models with available training data are updated. Hence, large amounts of training data are required in order to have significant recognition improvements. In some others, such as MLLR, where several general transformations are applied to model clusters, the results ar...
متن کامل