language model 0.00425747
language information 0.003036776
language models 0.003021508
model features 0.00300629
main language 0.002772473
factored language 0.0027687890000000002
language modeling 0.0027421800000000003
network language 0.0026663420000000004
matrix language 0.0026576950000000003
tional language 0.0025964710000000004
next language 0.002595041
combined language 0.0025739810000000004
embedded language 0.0025563960000000003
language identifier 0.0025317110000000003
language identifiers 0.0025284620000000004
language islands 0.002527637
first model 0.002411277
second model 0.002345044
model weight 0.002343682
ppl model 0.002306878
language 0.00227579
model this 0.002274533
flm model 0.002263877
model 0.00198168
previous word 0.0019188539999999998
current word 0.0019068849999999999
word form 0.0018767789999999999
next word 0.001837201
english words 0.00182329
rent word 0.0018222779999999999
word prediction 0.0017843899999999998
different feature 0.001721928
mandarin words 0.0017056559999999998
foreign words 0.001679828
vocabulary words 0.0016738859999999999
training data 0.0016731609999999998
embedded words 0.001646036
data corpus 0.001600686
speech corpus 0.001585604
different models 0.0015465140000000001
linguistic features 0.001500762
additional features 0.0014440310000000001
information tags 0.0014371509999999998
test set 0.0014228140000000001
different results 0.001383553
words 0.00136543
speech recognition 0.001364174
different time 0.001359655
general features 0.001357631
various features 0.001326125
previous feature 0.001322036
cal features 0.001320321
syntactical features 0.001309456
different backoff 0.001303436
multilingual speech 0.001298464
different smoothing 0.0012867619999999999
speech processing 0.0012821529999999999
monolingual data 0.001270598
audio data 0.001268427
set baseline 0.0012587990000000001
speech recogni 0.0012384079999999999
text corpus 0.001219123
output output 0.001209234
feature combinations 0.001196407
evaluation set 0.001175455
pos tags 0.001159467
output layer 0.0011382710000000002
different languages 0.001135984
different methods 0.0011004
text pos 0.00109896
different estimates 0.00108569
different topics 0.001069238
english output 0.001062477
several time 0.001058935
different kinds 0.00105693
different heuristics 0.001055167
different mono 0.001055167
guage models 0.001034632
ferent models 0.001025296
features 0.00102461
development set 0.0010038
speech 9.82139E-4
recognition system 9.81714E-4
other languages 9.76724E-4
set dev 9.72067E-4
bilingual training 9.69583E-4
train set 9.681959999999999E-4
other levels 9.60044E-4
acoustic modeling 9.58972E-4
set und 9.556269999999999E-4
velopment set 9.556269999999999E-4
input layer 9.44859E-4
other hand 9.34767E-4
wrong tags 9.29415E-4
much context 9.26543E-4
ing factors 9.246300000000001E-4
feature 9.21132E-4
hidden layer 9.176550000000001E-4
bilingual text 9.093010000000001E-4
other cases 9.09268E-4
