language model 0.0036579
language models 0.00263676
different word 0.00259724
model system 0.002544616
baseline model 0.002377218
acoustic model 0.002347955
word error 0.00226905
markov model 0.002261887
baseline language 0.002249078
word accuracy 0.002242844
based model 0.00220461
previous word 0.00219175
language modeling 0.002186952
language tasks 0.002169284
switchboard model 0.002156293
guage model 0.002153176
affine model 0.002151647
rate word 0.002120902
natural language 0.002080602
based language 0.00207647
word distributions 0.002074084
total word 0.00206798
tional language 0.002056968
word count 0.0020546229999999998
traditional language 0.002054056
word pairs 0.002028996
switchboard language 0.002028153
language processing 0.002024053
interpolated language 0.0020202609999999998
sri language 0.002013643
context feature 0.002011859
word insertions 0.002008188
word accu 0.002008188
context features 0.002001189
language users 0.001997614
text features 0.001938179
model 0.00189302
such features 0.001863677
translation models 0.001774687
language 0.00176488
feature vector 0.001683846
level features 0.0016011480000000002
video corpus 0.001588112
video information 0.001584742
pattern features 0.001540194
frequency words 0.0015308989999999998
motion features 0.001520955
event features 0.001517872
training corpus 0.001516927
feature streams 0.0015128
individual features 0.00150329
feature extraction 0.001503171
tern feature 0.0014959489999999999
informative features 0.001495938
such models 0.001485557
tent words 0.001473832
stop words 0.0014708009999999999
training data 0.0014471520000000002
large corpus 0.001424832
data set 0.001362401
baseline models 0.001356078
such video 0.001342985
acoustic models 0.001326815
machine translation 0.001319738
same speech 0.001285636
feature 0.00126067
features 0.00125
acoustic data 0.001243964
markov models 0.001240747
training set 0.001231495
words 0.0012203
temporal data 0.001219528
test set 0.001213351
different camera 0.00119827
paired corpus 0.001186962
bayesian models 0.001181477
information retrieval 0.00117343
text baseline 0.001172377
different lan 0.001168257
visual information 0.001147393
trigram models 0.001146803
large set 0.0011394
baseline system 0.0011357939999999999
initial results 0.001134253
guage models 0.001132036
speech recognition 0.001123189
switchboard corpus 0.001122077
paired data 0.001117187
linguistic context 0.001109828
models condition 0.001104728
vide information 0.001101705
different television 0.001095782
parallel data 0.001091117
ful information 0.001090484
leverage information 0.001090484
precise information 0.001090484
different transcriptions 0.001089894
sports video 0.001085
positive results 0.001073655
different directors 0.001057905
