Title :
Large vocabulary continuous speech recognition using HTK
Author :
Woodland, P.C. ; Odell, J.J. ; Valtchev, V. ; Young, S.J.
Author_Institution :
Dept. of Eng., Cambridge Univ., UK
Abstract :
HTK is a portable software toolkit for building speech recognition systems using continuous density hidden Markov models developed by the Cambridge University Speech Group. One particularly successful type of system uses mixture density tied-state triphones. We have used this technique for the 5 k/20 k word ARPA Wall Street Journal (WSJ) task. We have extended our approach from using word-internal gender independent modelling to use decision tree based state clustering, cross-word triphones and gender dependent models. Our current systems can be run with either bigram or trigram language models using a single pass dynamic network decoder. Systems based on these techniques were included in the November 1993 ARPA WSJ evaluation, and gave the lowest error rate reported on the 5 k word bigram, 5 k word trigram and 20 k word bigram “hub” tests and the second lowest error rate on the 20 k word trigram “hub” test
Keywords :
decision theory; decoding; hidden Markov models; natural languages; software tools; speech recognition; vocabulary; ARPA Wall Street Journal task; Cambridge University Speech Group; HTK; bigram language models; continuous density hidden Markov models; continuous speech recognition; cross-word triphones; decision tree based state clustering; error rate; gender dependent models; mixture density tied-state triphones; portable software toolkit; single pass dynamic network decoder; speech recognition systems; trigram language models; word-internal gender independent modelling; Buildings; Decision trees; Decoding; Error analysis; Hidden Markov models; Software tools; Speech recognition; System testing; Training data; Vocabulary;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1994. ICASSP-94., 1994 IEEE International Conference on
Conference_Location :
Adelaide, SA
Print_ISBN :
0-7803-1775-0
DOI :
10.1109/ICASSP.1994.389562