원문정보
초록
영어
In speaker Identification systems both parametric and nonparametric probability modeling is used. The Gaussian model is the basic parametric model that is used and this model is the basis of other sophisticated and it can be performed in a completely text independent situation. However, it sounds efficient to speaker identification application, but it results long time processing in practice. In this paper, we propose a decision function by using vector quantization (VQ) techniques to decrease the training model for GMM in order to reduce the processing time. In our proposed modeling, we take the superiority of VQ, which is simplicity computation to distinguish between male and female speaker. Then, GMM is applied into the subgroup of speaker to get the accuracy rates. Experimental result shows that our hybrid VQ/GMM method always yielded better improvements in accuracy and bring reduce in time processing. All the experiments have been done in both direct recording speech and mobile phone speech signals.
목차
1. Introduction
2. Previous Works
2.1 Feature Extraction
3. Vector Quantization
3.1. Training Model Based On Clustering Technique:
4. Mixture Model
4.1. Gaussian Mixture Model:
5. Code Separated GMM:
7. Experimental Setup
8. Results
9. Conclusion
10. Summary
References