A TIME-DELAY NEURAL NETWORK ARCHITECTURE FOR ISOLATED WORD RECOGNITION

被引:255
作者
LANG, KJ
WAIBEL, AH
HINTON, GE
机构
[1] CARNEGIE MELLON UNIV,PITTSBURGH,PA 15213
[2] UNIV TORONTO,TORONTO M5S 1A1,ONTARIO,CANADA
关键词
Constrained links; Isolated word recognition; Multiresolution learning; Multispeaker speech recognition; Network architecture; Neural networks; Time delays;
D O I
10.1016/0893-6080(90)90044-L
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
A translation-invariant back-propagation network is described that performs better than a sophisticated continuous acoustic parameter hidden Markov model on a noisy, 100-speaker confusable vocabulary isolated word recognition task. The network's replicated architecture permits it to extract precise information from unaligned training patterns selected by a naive segmentation rule. © 1990.
引用
收藏
页码:23 / 43
页数:21
相关论文
共 21 条
[1]   A MAXIMUM-LIKELIHOOD APPROACH TO CONTINUOUS SPEECH RECOGNITION [J].
BAHL, LR ;
JELINEK, F ;
MERCER, RL .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1983, 5 (02) :179-190
[2]  
BAKER JK, 1975, SPEECH RECOGNITION, P521
[3]  
BROWN PF, 1987, THESIS CARNEGIE MELL
[4]  
Duda R. O., 1973, PATTERN CLASSIFICATI, V3
[6]  
HINTON GE, 1987, CMUCS87115 CARN MELL
[7]  
HINTON GE, 1987, 9TH P ANN C COGN SCI
[8]  
HINTON GE, 1987, PARLE PARALLEL ARCHI
[9]  
JORDAN MI, 1986, 8TH P ANN C COGN SCI
[10]  
LANG KJ, 1987, THESIS CARNEGIE MELL