word test 0.002773681
pronunciation model 0.0026109790000000003
single word 0.002548007
discriminative model 0.002543213
articulatory model 0.002534356
word sequences 0.002528782
linear model 0.002513008
generative model 0.002429832
word score 0.002426619
alignment feature 0.002426106
probabilistic model 0.002411099
dbn model 0.002408686
feature set 0.002389763
word units 0.002386733
model com 0.002385019
corresponding word 0.002367306
word tokens 0.002352035
dicted word 0.002339128
confusable word 0.002339128
word discrimination 0.002339128
word queries 0.002339128
dictionary feature 0.00225064
different feature 0.002150603
model 0.00212882
feature function 0.0020656340000000002
length feature 0.002036951
articulatory feature 0.001997036
new feature 0.001992916
other words 0.001951864
final feature 0.001933819
feature selection 0.0019182930000000002
feature space 0.0019049330000000001
feature functions 0.001873197
feature sets 0.001858834
each feature 0.001857976
data set 0.001853683
substitution feature 0.001848986
feature func 0.001841073
alignment features 0.0018306120000000001
common words 0.00182098
speech corpus 0.00172076
training set 0.0017043639999999999
same data 0.001671021
few words 0.001640888
frequent words 0.001627767
same features 0.0016116070000000001
feature 0.0015915
different algorithm 0.0015004900000000002
ing data 0.001486136
test set 0.001479264
phonetic alignment 0.001464447
length features 0.001441457
articulatory features 0.001401542
learning algorithm 0.001399338
first set 0.001392291
words 0.00137564
large set 0.001373287
conversational data 0.001366154
phone set 0.001350981
data subset 0.0013222470000000001
levenshtein features 0.001299445
dbn features 0.001275872
set performance 0.001267492
latory features 0.001262458
asynchrony features 0.001260473
ment features 0.001257561
substitution features 0.0012534920000000001
ticulatory features 0.001252158
training time 0.001248102
ing set 0.001228979
phonetic error 0.001207602
online algorithm 0.001203703
such models 0.001199506
pegasos algorithm 0.001197297
programming alignment 0.0011888530000000001
algorithm corre 0.001188177
alignment score 0.001168545
speech recognition 0.00116664
conversational speech 0.001161937
good alignment 0.001158957
pronunciation models 0.001156546
alignment fea 0.001149042
pronunciation dictionary 0.0011412990000000001
netic alignment 0.001131351
dbn alignment 0.001114472
development set 0.001113113
speech recognizers 0.0011043890000000001
automatic speech 0.001102965
speech recog 0.001102502
complete speech 0.001100029
speech database 0.001098561
discriminative models 0.0010887800000000001
same performance 0.00108483
articulatory models 0.001079923
same length 0.001061052
predefined set 0.001044412
other length 0.0010216749999999999
error rate 0.001003476
test examples 9.99846E-4
features 9.96006E-4
