Title :
Efficient decoding algorithms for Mandarin connected digit speech recognition
Author :
Xuan, Zhu ; Husheng, Li ; Jia, Liu ; Runsheng, Liu
Author_Institution :
Dept. of Electron. Eng., Tsinghua Univ., Beijing, China
Abstract :
In this paper, a set of efficient decoding algorithms based on a speaker independent Mandarin connected digit speech recognition system is proposed. By simplifying the computation of observation probabilities and adopting an improved beam search pruning algorithm combined with duration information, the average decoding time is reduced from 0.92 second to 0.11 second per digit string while the recognition rate of unknown-length string is only decreased from 94.3% to 94.0%
Keywords :
decoding; search problems; speech coding; speech recognition; Mandarin connected digit speech recognition; beam search pruning algorithm; decoding algorithms; observation probabilities; speaker independent recognition; unknown-length string; Automatic speech recognition; Decoding; Gaussian distribution; Hidden Markov models; Mel frequency cepstral coefficient; Probability; Psychology; Real time systems; Speech recognition; Viterbi algorithm;
Conference_Titel :
Intelligent Multimedia, Video and Speech Processing, 2001. Proceedings of 2001 International Symposium on
Conference_Location :
Hong Kong
Print_ISBN :
962-85766-2-3
DOI :
10.1109/ISIMP.2001.925456
