Title :
State duration-based segmental probability model
Author :
Jia, Bin ; Zhu, Xiaoyan ; Luo, Yuping ; Hu, Dongcheng
Author_Institution :
Dept. of Autom., Tsinghua Univ., Beijing, China
Abstract :
This paper suggests a state duration-based segmental probability model (SDSPM) for speech recognition. It comes from incorporating the concept of state duration into the SPM. The distribution of state duration is represented by the bounded gamma distribution (BGD), considered to be better than the gamma distribution. The SDSPM is simpler than the HMM. But the experiments show an average 6% improvement of the rate of recognition accuracy compared with HMM and SPM
Keywords :
gamma distribution; speech recognition; bounded gamma distribution; speech recognition; state duration-based segmental probability model; Automation; Computational efficiency; Computer science; Hidden Markov models; Intelligent systems; Laboratories; Probability; Scanning probe microscopy; Speech recognition; Stochastic processes;
Conference_Titel :
Communication Technology Proceedings, 1998. ICCT '98. 1998 International Conference on
Conference_Location :
Beijing
Print_ISBN :
7-80090-827-5
DOI :
10.1109/ICCT.1998.741434
