Island-driven continuous speech recognizer using phone-based HMM word spotting
- 13 January 2003
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
Abstract
An island-driven speech recognition system using word-spotting and word-verification processes based on HMM (hidden Markov model) phone units is proposed. First, the system detects a keyword as an island in continuous speech. After that, the system expands the island by verifying neighbor words. Word models for detection or verification are composed of HMM phone units. The proposed speech recognition system was tested by word-spotting and phrase-recognition experiments using 75 sentences which contained 549 phrases uttered by three male speakers. The noun-detection rate is 96.5% with 140 false alarms (in 549 phrases). The phrase-recognition rate is 96.2% up to the three best candidates.Keywords
This publication has 4 references indexed in Scilit:
- BYBLOS: The BBN continuous speech recognition systemPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- High performance connected digit recognition, using hidden Markov modelsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- Large-vocabulary speaker-independent continuous speech recognition using HMMPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- LPC peak weighted spectral matching measuresElectronics and Communications in Japan (Part I: Communications), 1981