DocumentCode
310458
Title
Speech recognition using neural networks with forward-backward probability generated targets
Author
Yonghong Yen ; Fanty, Mark ; Cole, Ron
Author_Institution
Center for Spoken Language Understanding, Oregon Graduate Inst. of Sci. & Technol., Portland, OR, USA
Volume
4
fYear
1997
fDate
21-24 Apr 1997
Firstpage
3241
Abstract
Neural network training targets for speech recognition are estimated using a novel method. Rather than use zero and one, continuous targets are generated using forward-backward probabilities. Each training pattern has more than one class active. Experiments showed that the new method effectively decreased the error rate by 15% in a continuous digits recognition task
Keywords
hidden Markov models; learning (artificial intelligence); neural nets; probability; speech processing; speech recognition; HMM; continuous digits recognition task; continuous targets; error rate reduction; experiments; forward-backward probability generated targets; neural networks; speech recognition; training pattern; training targets; Databases; Decoding; Error analysis; Hidden Markov models; Natural languages; Neural networks; Speech analysis; Speech recognition; Training data; Viterbi algorithm;
fLanguage
English
Publisher
ieee
Conference_Titel
Acoustics, Speech, and Signal Processing, 1997. ICASSP-97., 1997 IEEE International Conference on
Conference_Location
Munich
ISSN
1520-6149
Print_ISBN
0-8186-7919-0
Type
conf
DOI
10.1109/ICASSP.1997.595483
Filename
595483
Link To Document