asr word 0.00364426
asr asr 0.00329148
model asr 0.0029379799999999998
word error 0.002906444
other words 0.002669883
asr error 0.0025536639999999998
word confidence 0.002546088
training data 0.002525701
simple word 0.002419245
previous word 0.002412818
word segmentation 0.0023997330000000002
models asr 0.002398336
word pos 0.0023881970000000003
other features 0.002368523
noisy word 0.002354647
speech data 0.002352895
ner model 0.00234809
word accuracy 0.002343937
word posterior 0.002322801
word hypotheses 0.002306803
word murayama 0.002290746
erroneous word 0.0022744740000000003
word sequences 0.0022596300000000003
word lattices 0.002257709
word durations 0.0022533130000000003
asr system 0.002251871
word graphs 0.0022505000000000003
word poste 0.0022505000000000003
asr results 0.002245488
asr errors 0.002217816
asr confidence 0.0021933079999999997
prosodic features 0.002181251
character words 0.002153267
context words 0.002147838
ner ner 0.0021117
many words 0.002087819
text data 0.0020827930000000003
feature values 0.002079799
confidence feature 0.002044368
training ner 0.002043331
many asr 0.002033419
nized words 0.001994447
asr accuracy 0.001991157
language model 0.001972766
transcription asr 0.001961286
misrecognized words 0.00195766
asr applications 0.001955392
asr hypotheses 0.001954023
unconfident words 0.001953953
ognized words 0.001953953
transcriptions asr 0.001949723
effective asr 0.0019387739999999999
multiple asr 0.0019321859999999998
same model 0.00192683
asr hypothe 0.001917237
asr con 0.001912885
perfect asr 0.001909671
pass asr 0.0019071889999999999
asr confi 0.0019005189999999998
asr engine 0.001898442
rating asr 0.0018971449999999998
ing features 0.0018893220000000001
feature value 0.001867879
ner performance 0.001839218
ner models 0.001808446
feature selection 0.001805246
binary features 0.001804979
dence feature 0.0017556989999999999
feature indi 0.001753642
fidence feature 0.001749788
dominant feature 0.001749788
type features 0.001746397
model score 0.001715942
words 0.00170014
model figure 0.001698085
ple features 0.0016937270000000001
acoustic model 0.0016781
ner results 0.001655598
tional features 0.0016555570000000002
tiple features 0.00165177
model weight 0.00162984
training test 0.0015858019999999999
guage model 0.001559212
confidence training 0.001535049
training set 0.001525989
other approach 0.001513363
feature 0.0014968
ner method 0.0014900450000000002
nes ner 0.001484302
ner recall 0.0014423600000000002
ner scores 0.001429279
ner level 0.001401939
features 0.00139878
other tasks 0.001386506
ner precision 0.001375645
ner perfor 0.0013729620000000001
different kinds 0.001362672
ner experiments 0.0013572200000000001
discriminative ner 0.0013497580000000002
good ner 0.001339609
