원문정보
보안공학연구지원센터(IJSEIA)
International Journal of Software Engineering and Its Applications
Vol.10 No.1
2016.01
pp.279-286
피인용수 : 0건 (자료제공 : 네이버학술정보)
초록
영어
This paper proposes a new feature vector selection method for voice pattern recognition tasks, especially for speaker or emotion recognition. During the model training phase, robust speaker or emotion models are constructed by using meaningful feature vectors while discarding confusing vectors that may induce recognition error. To select meaningful feature vectors, the proposed method classifies feature vectors into overlapped and non-overlapped sets using log-likelihood ratio. Speaker- and emotion-recognition experiments confirmed that these robust models significantly reduce recognition errors.
목차
Abstract
1. Introduction
2. The Conventional GMM-based Pattern Recognition Technique
2.1. Principle of GMM-based Speaker and Emotion Recognition
2.2. Feature Vector Selection
3. Feature Vector Selection based on Log-Likelihood
4. Experiments and Results
4.1. Speaker Recognition Experiments
4.2. Emotion Recognition Experiments
5. Conclusions
References
1. Introduction
2. The Conventional GMM-based Pattern Recognition Technique
2.1. Principle of GMM-based Speaker and Emotion Recognition
2.2. Feature Vector Selection
3. Feature Vector Selection based on Log-Likelihood
4. Experiments and Results
4.1. Speaker Recognition Experiments
4.2. Emotion Recognition Experiments
5. Conclusions
References
저자정보
참고문헌
자료제공 : 네이버학술정보