The Reserch And Application Of Text-Dependent Speaker Recognition Technology

In text-dependent speaker recognition system, usually defines the contents of the speech, thereby limiting its application scenarios. By studying speaker identification algorithm based on the difference between different phonemes statistical features and phonemes in speech research-based, reference other papers related algorithm, propose for the short-time speech robustness based the phoneme segmentation speaker recognition system design program, and to be achieved.Firstly speaker identification system feature extraction and modeling methods and speech recognition technology to be introduced. Completed on the basis of learning and understanding the theory and completed on the Sphinx series software code analysis, such as Sphinxbase, Sphinxtrain as well as Sphinx-4. And complete phonemic segments based on Sphinx-4decoder module.Then describes commonly used algorithm of text-dependent speaker recognition system, followed by the introduction of two phoneme-based speaker recognition system, and completed the design of text-dependent speaker identification system with related algorithm in two system above. Completed programming on recognition module, and realized the speaker recognition system based on software analysis for Sphinx series open source project.Finally, do the analysis of the statistical properties of the phonemes and phoneme cluster, and evaluate phoneme segmentation results combined with the clustering results, and to explore its impact on the identification module. Finally, do experiments between this speaker recognition system with other two system based on phonemes, and show out the experimental and analytical results.